Authors: Rajeev Goel, Utkarsh Nath, Yancheng Wang, Alvin C. Silva, Teresa Wu, Yingzhen Yang
Abstract: Deep neural networks, including Convolutional Neural Networks (CNNs) and Visual Transformers (ViT), have achieved stunning success in medical image domain. We study thorax disease classification in this paper. Effective extraction of features for the disease areas is crucial for disease classification on radiographic images. While various neural architectures and training techniques, such as self-supervised learning with contrastive/restorative learning, have been employed for disease classification on radiographic images, there are no principled methods which can effectively reduce the adverse effect of noise and background, or non-disease areas, on the radiographic images for disease classification. To address this challenge, we propose a novel Low-Rank Feature Learning (LRFL) method in this paper, which is universally applicable to the training of all neural networks. The LRFL method is both empirically motivated by the low frequency property observed on all the medical datasets in this paper, and theoretically motivated by our sharp generalization bound for neural networks with low-rank features. In the empirical study, using a neural network such as a ViT or a CNN pre-trained on unlabeled chest X-rays by Masked Autoencoders (MAE), our novel LRFL method is applied on the pre-trained neural network and demonstrate better classification results in terms of both multiclass area under the receiver operating curve (mAUC) and classification accuracy.
Authors: Michelle R. Greene, Benjamin J. Balas, Mark D. Lescroart, Paul R. MacNeilage, Jennifer A. Hart, Kamran Binaee, Peter A. Hausamann, Ronald Mezile, Bharath Shankar, Christian B. Sinnott, Kaylie Capurro, Savannah Halow, Hunter Howe, Mariam Josyula, Annie Li, Abraham Mieses, Amina Mohamed, Ilya Nudnou, Ezra Parkhill, Peter Riley, Brett Schmidt, Matthew W. Shinkle, Wentao Si, Brian Szekely, Joaquin M. Torres, Eliana Weissmann
Abstract: We introduce the Visual Experience Dataset (VEDB), a compilation of over 240 hours of egocentric video combined with gaze- and head-tracking data that offers an unprecedented view of the visual world as experienced by human observers. The dataset consists of 717 sessions, recorded by 58 observers ranging from 6-49 years old. This paper outlines the data collection, processing, and labeling protocols undertaken to ensure a representative sample and discusses the potential sources of error or bias within the dataset. The VEDB's potential applications are vast, including improving gaze tracking methodologies, assessing spatiotemporal image statistics, and refining deep neural networks for scene and activity recognition. The VEDB is accessible through established open science platforms and is intended to be a living dataset with plans for expansion and community contributions. It is released with an emphasis on ethical considerations, such as participant privacy and the mitigation of potential biases. By providing a dataset grounded in real-world experiences and accompanied by extensive metadata and supporting code, the authors invite the research community to utilize and contribute to the VEDB, facilitating a richer understanding of visual perception and behavior in naturalistic settings.
Authors: Sourabh Vasant Gothe, Vibhav Agarwal, Sourav Ghosh, Jayesh Rajkumar Vachhani, Pranay Kashyap, Barath Raj Kandur Raja
Abstract: Generic Event Boundary Detection (GEBD) task aims to recognize generic, taxonomy-free boundaries that segment a video into meaningful events. Current methods typically involve a neural model trained on a large volume of data, demanding substantial computational power and storage space. We explore two pivotal questions pertaining to GEBD: Can non-parametric algorithms outperform unsupervised neural methods? Does motion information alone suffice for high performance? This inquiry drives us to algorithmically harness motion cues for identifying generic event boundaries in videos. In this work, we propose FlowGEBD, a non-parametric, unsupervised technique for GEBD. Our approach entails two algorithms utilizing optical flow: (i) Pixel Tracking and (ii) Flow Normalization. By conducting thorough experimentation on the challenging Kinetics-GEBD and TAPOS datasets, our results establish FlowGEBD as the new state-of-the-art (SOTA) among unsupervised methods. FlowGEBD exceeds the neural models on the Kinetics-GEBD dataset by obtaining an F1@0.05 score of 0.713 with an absolute gain of 31.7% compared to the unsupervised baseline and achieves an average F1 score of 0.623 on the TAPOS validation dataset.
Authors: Damith Chamalke Senadeera, Xiaoyun Yang, Dimitrios Kollias, Gregory Slabaugh
Abstract: In this paper we introduce CUE-Net, a novel architecture designed for automated violence detection in video surveillance. As surveillance systems become more prevalent due to technological advances and decreasing costs, the challenge of efficiently monitoring vast amounts of video data has intensified. CUE-Net addresses this challenge by combining spatial Cropping with an enhanced version of the UniformerV2 architecture, integrating convolutional and self-attention mechanisms alongside a novel Modified Efficient Additive Attention mechanism (which reduces the quadratic time complexity of self-attention) to effectively and efficiently identify violent activities. This approach aims to overcome traditional challenges such as capturing distant or partially obscured subjects within video frames. By focusing on both local and global spatiotemporal features, CUE-Net achieves state-of-the-art performance on the RWF-2000 and RLVS datasets, surpassing existing methods.
Authors: Yuan Wang, Huazhu Fu, Renuga Kanagavelu, Qingsong Wei, Yong Liu, Rick Siow Mong Goh
Abstract: The performance of Federated Learning (FL) hinges on the effectiveness of utilizing knowledge from distributed datasets. Traditional FL methods adopt an aggregate-then-adapt framework, where clients update local models based on a global model aggregated by the server from the previous training round. This process can cause client drift, especially with significant cross-client data heterogeneity, impacting model performance and convergence of the FL algorithm. To address these challenges, we introduce FedAF, a novel aggregation-free FL algorithm. In this framework, clients collaboratively learn condensed data by leveraging peer knowledge, the server subsequently trains the global model using the condensed data and soft labels received from the clients. FedAF inherently avoids the issue of client drift, enhances the quality of condensed data amid notable data heterogeneity, and improves the global model performance. Extensive numerical studies on several popular benchmark datasets show FedAF surpasses various state-of-the-art FL algorithms in handling label-skew and feature-skew data heterogeneity, leading to superior global model accuracy and faster convergence.
Authors: Nagabhushan Somraj, Adithyan Karanayil, Sai Harsha Mupparaju, Rajiv Soundararajan
Abstract: Neural Radiance Fields (NeRF) show impressive performance in photo-realistic free-view rendering of scenes. Recent improvements on the NeRF such as TensoRF and ZipNeRF employ explicit models for faster optimization and rendering, as compared to the NeRF that employs an implicit representation. However, both implicit and explicit radiance fields require dense sampling of images in the given scene. Their performance degrades significantly when only a sparse set of views is available. Researchers find that supervising the depth estimated by a radiance field helps train it effectively with fewer views. The depth supervision is obtained either using classical approaches or neural networks pre-trained on a large dataset. While the former may provide only sparse supervision, the latter may suffer from generalization issues. As opposed to the earlier approaches, we seek to learn the depth supervision by designing augmented models and training them along with the main radiance field. Further, we aim to design a framework of regularizations that can work across different implicit and explicit radiance fields. We observe that certain features of these radiance field models overfit to the observed images in the sparse-input scenario. Our key finding is that reducing the capability of the radiance fields with respect to positional encoding, the number of decomposed tensor components or the size of the hash table, constrains the model to learn simpler solutions, which estimate better depth in certain regions. By designing augmented models based on such reduced capabilities, we obtain better depth supervision for the main radiance field. We achieve state-of-the-art view-synthesis performance with sparse input views on popular datasets containing forward-facing and 360$^\circ$ scenes by employing the above regularizations.
Authors: Lei Kang, Rub\`en Tito, Ernest Valveny, Dimosthenis Karatzas
Abstract: Documents are 2-dimensional carriers of written communication, and as such their interpretation requires a multi-modal approach where textual and visual information are efficiently combined. Document Visual Question Answering (Document VQA), due to this multi-modal nature, has garnered significant interest from both the document understanding and natural language processing communities. The state-of-the-art single-page Document VQA methods show impressive performance, yet in multi-page scenarios, these methods struggle. They have to concatenate all pages into one large page for processing, demanding substantial GPU resources, even for evaluation. In this work, we propose a novel method and efficient training strategy for multi-page Document VQA tasks. In particular, we employ a visual-only document representation, leveraging the encoder from a document understanding model, Pix2Struct. Our approach utilizes a self-attention scoring mechanism to generate relevance scores for each document page, enabling the retrieval of pertinent pages. This adaptation allows us to extend single-page Document VQA models to multi-page scenarios without constraints on the number of pages during evaluation, all with minimal demand for GPU resources. Our extensive experiments demonstrate not only achieving state-of-the-art performance without the need for Optical Character Recognition (OCR), but also sustained performance in scenarios extending to documents of nearly 800 pages compared to a maximum of 20 pages in the MP-DocVQA dataset. Our code is publicly available at \url{https://github.com/leitro/SelfAttnScoring-MPDocVQA}.
Authors: Cong Wang, Di Kang, He-Yi Sun, Shen-Han Qian, Zi-Xuan Wang, Linchao Bao, Song-Hai Zhang
Abstract: Creating high-fidelity head avatars from multi-view videos is a core issue for many AR/VR applications. However, existing methods usually struggle to obtain high-quality renderings for all different head components simultaneously since they use one single representation to model components with drastically different characteristics (e.g., skin vs. hair). In this paper, we propose a Hybrid Mesh-Gaussian Head Avatar (MeGA) that models different head components with more suitable representations. Specifically, we select an enhanced FLAME mesh as our facial representation and predict a UV displacement map to provide per-vertex offsets for improved personalized geometric details. To achieve photorealistic renderings, we obtain facial colors using deferred neural rendering and disentangle neural textures into three meaningful parts. For hair modeling, we first build a static canonical hair using 3D Gaussian Splatting. A rigid transformation and an MLP-based deformation field are further applied to handle complex dynamic expressions. Combined with our occlusion-aware blending, MeGA generates higher-fidelity renderings for the whole head and naturally supports more downstream tasks. Experiments on the NeRSemble dataset demonstrate the effectiveness of our designs, outperforming previous state-of-the-art methods and supporting various editing functionalities, including hairstyle alteration and texture editing.
Authors: Lei Kang, Mohamed Ali Souibgui, Fei Yang, Lluis Gomez, Ernest Valveny, Dimosthenis Karatzas
Abstract: Document understanding models have recently demonstrated remarkable performance by leveraging extensive collections of user documents. However, since documents often contain large amounts of personal data, their usage can pose a threat to user privacy and weaken the bonds of trust between humans and AI services. In response to these concerns, legislation advocating ``the right to be forgotten" has recently been proposed, allowing users to request the removal of private information from computer systems and neural network models. A novel approach, known as machine unlearning, has emerged to make AI models forget about a particular class of data. In our research, we explore machine unlearning for document classification problems, representing, to the best of our knowledge, the first investigation into this area. Specifically, we consider a realistic scenario where a remote server houses a well-trained model and possesses only a small portion of training data. This setup is designed for efficient forgetting manipulation. This work represents a pioneering step towards the development of machine unlearning methods aimed at addressing privacy concerns in document analysis applications. Our code is publicly available at \url{https://github.com/leitro/MachineUnlearning-DocClassification}.
URLs: https://github.com/leitro/MachineUnlearning-DocClassification
Authors: Tianyong Wang, Xiangyu Liang, Wangguandong Zheng, Dan Niu, Haifeng Xia, Siyu Xia
Abstract: The talking head generation recently attracted considerable attention due to its widespread application prospects, especially for digital avatars and 3D animation design. Inspired by this practical demand, several works explored Neural Radiance Fields (NeRF) to synthesize the talking heads. However, these methods based on NeRF face two challenges: (1) Difficulty in generating style-controllable talking heads. (2) Displacement artifacts around the neck in rendered images. To overcome these two challenges, we propose a novel generative paradigm \textit{Embedded Representation Learning Network} (ERLNet) with two learning stages. First, the \textit{ audio-driven FLAME} (ADF) module is constructed to produce facial expression and head pose sequences synchronized with content audio and style video. Second, given the sequence deduced by the ADF, one novel \textit{dual-branch fusion NeRF} (DBF-NeRF) explores these contents to render the final images. Extensive empirical studies demonstrate that the collaboration of these two stages effectively facilitates our method to render a more realistic talking head than the existing algorithms.
Authors: Bo Chen, Shoukang Hu, Qi Chen, Chenpeng Du, Ran Yi, Yanmin Qian, Xie Chen
Abstract: We present GStalker, a 3D audio-driven talking face generation model with Gaussian Splatting for both fast training (40 minutes) and real-time rendering (125 FPS) with a 3$\sim$5 minute video for training material, in comparison with previous 2D and 3D NeRF-based modeling frameworks which require hours of training and seconds of rendering per frame. Specifically, GSTalker learns an audio-driven Gaussian deformation field to translate and transform 3D Gaussians to synchronize with audio information, in which multi-resolution hashing grid-based tri-plane and temporal smooth module are incorporated to learn accurate deformation for fine-grained facial details. In addition, a pose-conditioned deformation field is designed to model the stabilized torso. To enable efficient optimization of the condition Gaussian deformation field, we initialize 3D Gaussians by learning a coarse static Gaussian representation. Extensive experiments in person-specific videos with audio tracks validate that GSTalker can generate high-fidelity and audio-lips synchronized results with fast training and real-time rendering speed.
Authors: Hyunho Lee, Wenwen Li
Abstract: Flood inundation mapping is a critical task for responding to the increasing risk of flooding linked to global warming. Significant advancements of deep learning in recent years have triggered its extensive applications, including flood inundation mapping. To cope with the time-consuming and labor-intensive data labeling process in supervised learning, deep active learning strategies are one of the feasible approaches. However, there remains limited exploration into the interpretability of how deep active learning strategies operate, with a specific focus on flood inundation mapping in the field of remote sensing. In this study, we introduce a novel framework of Interpretable Deep Active Learning for Flood inundation Mapping (IDAL-FIM), specifically in terms of class ambiguity of multi-spectral satellite images. In the experiments, we utilize Sen1Floods11 dataset, and adopt U-Net with MC-dropout. In addition, we employ five acquisition functions, which are the random, K-means, BALD, entropy, and margin acquisition functions. Based on the experimental results, we demonstrate that two proposed class ambiguity indices are effective variables to interpret the deep active learning by establishing statistically significant correlation with the predictive uncertainty of the deep learning model at the tile level. Then, we illustrate the behaviors of deep active learning through visualizing two-dimensional density plots and providing interpretations regarding the operation of deep active learning, in flood inundation mapping.
Authors: Susano Mingwin, Yulong Shisu, Yongshuai Wanwag, Sunshin Huing
Abstract: This research introduces an innovative method for Traffic Sign Recognition (TSR) by leveraging deep learning techniques, with a particular emphasis on Vision Transformers. TSR holds a vital role in advancing driver assistance systems and autonomous vehicles. Traditional TSR approaches, reliant on manual feature extraction, have proven to be labor-intensive and costly. Moreover, methods based on shape and color have inherent limitations, including susceptibility to various factors and changes in lighting conditions. This study explores three variants of Vision Transformers (PVT, TNT, LNL) and six convolutional neural networks (AlexNet, ResNet, VGG16, MobileNet, EfficientNet, GoogleNet) as baseline models. To address the shortcomings of traditional methods, a novel pyramid EATFormer backbone is proposed, amalgamating Evolutionary Algorithms (EAs) with the Transformer architecture. The introduced EA-based Transformer block captures multi-scale, interactive, and individual information through its components: Feed-Forward Network, Global and Local Interaction, and Multi-Scale Region Aggregation modules. Furthermore, a Modulated Deformable MSA module is introduced to dynamically model irregular locations. Experimental evaluations on the GTSRB and BelgiumTS datasets demonstrate the efficacy of the proposed approach in enhancing both prediction speed and accuracy. This study concludes that Vision Transformers hold significant promise in traffic sign classification and contributes a fresh algorithmic framework for TSR. These findings set the stage for the development of precise and dependable TSR algorithms, benefiting driver assistance systems and autonomous vehicles.
Authors: Hongrui Zhao, Michael F. Lembeck, Adrian Zhuang, Riya Shah, Jesse Wei
Abstract: Star trackers are one of the most accurate celestial sensors used for absolute attitude determination. The devices detect stars in captured images and accurately compute their projected centroids on an imaging focal plane with subpixel precision. Traditional algorithms for star detection and centroiding often rely on threshold adjustments for star pixel detection and pixel brightness weighting for centroid computation. However, challenges like high sensor noise and stray light can compromise algorithm performance. This article introduces a Convolutional Neural Network (CNN)-based approach for star detection and centroiding, tailored to address the issues posed by noisy star tracker images in the presence of stray light and other artifacts. Trained using simulated star images overlayed with real sensor noise and stray light, the CNN produces both a binary segmentation map distinguishing star pixels from the background and a distance map indicating each pixel's proximity to the nearest star centroid. Leveraging this distance information alongside pixel coordinates transforms centroid calculations into a set of trilateration problems solvable via the least squares method. Our method employs efficient UNet variants for the underlying CNN architectures, and the variants' performances are evaluated. Comprehensive testing has been undertaken with synthetic image evaluations, hardware-in-the-loop assessments, and night sky tests. The tests consistently demonstrated that our method outperforms several existing algorithms in centroiding accuracy and exhibits superior resilience to high sensor noise and stray light interference. An additional benefit of our algorithms is that they can be executed in real-time on low-power edge AI processors.
Authors: Nikita Drobyshev, Antoni Bigata Casademunt, Konstantinos Vougioukas, Zoe Landgraf, Stavros Petridis, Maja Pantic
Abstract: Head avatars animated by visual signals have gained popularity, particularly in cross-driving synthesis where the driver differs from the animated character, a challenging but highly practical approach. The recently presented MegaPortraits model has demonstrated state-of-the-art results in this domain. We conduct a deep examination and evaluation of this model, with a particular focus on its latent space for facial expression descriptors, and uncover several limitations with its ability to express intense face motions. To address these limitations, we propose substantial changes in both training pipeline and model architecture, to introduce our EMOPortraits model, where we: Enhance the model's capability to faithfully support intense, asymmetric face expressions, setting a new state-of-the-art result in the emotion transfer task, surpassing previous methods in both metrics and quality. Incorporate speech-driven mode to our model, achieving top-tier performance in audio-driven facial animation, making it possible to drive source identity through diverse modalities, including visual signal, audio, or a blend of both. We propose a novel multi-view video dataset featuring a wide range of intense and asymmetric facial expressions, filling the gap with absence of such data in existing datasets.
Authors: Alexis Guichemerre, Soufiane Belharbi, Tsiry Mayet, Shakeeb Murtaza, Pourya Shamsolmoali, Luke McCaffrey, Eric Granger
Abstract: Given the emergence of deep learning, digital pathology has gained popularity for cancer diagnosis based on histology images. Deep weakly supervised object localization (WSOL) models can be trained to classify histology images according to cancer grade and identify regions of interest (ROIs) for interpretation, using inexpensive global image-class annotations. A WSOL model initially trained on some labeled source image data can be adapted using unlabeled target data in cases of significant domain shifts caused by variations in staining, scanners, and cancer type. In this paper, we focus on source-free (unsupervised) domain adaptation (SFDA), a challenging problem where a pre-trained source model is adapted to a new target domain without using any source domain data for privacy and efficiency reasons. SFDA of WSOL models raises several challenges in histology, most notably because they are not intended to adapt for both classification and localization tasks. In this paper, 4 state-of-the-art SFDA methods, each one representative of a main SFDA family, are compared for WSOL in terms of classification and localization accuracy. They are the SFDA-Distribution Estimation, Source HypOthesis Transfer, Cross-Domain Contrastive Learning, and Adaptively Domain Statistics Alignment. Experimental results on the challenging Glas (smaller, breast cancer) and Camelyon16 (larger, colon cancer) histology datasets indicate that these SFDA methods typically perform poorly for localization after adaptation when optimized for classification.
Authors: Christopher J. Kymn, Sonia Mazelet, Annabel Ng, Denis Kleyko, Bruno A. Olshausen
Abstract: We propose a system for visual scene analysis and recognition based on encoding the sparse, latent feature-representation of an image into a high-dimensional vector that is subsequently factorized to parse scene content. The sparse feature representation is learned from image statistics via convolutional sparse coding, while scene parsing is performed by a resonator network. The integration of sparse coding with the resonator network increases the capacity of distributed representations and reduces collisions in the combinatorial search space during factorization. We find that for this problem the resonator network is capable of fast and accurate vector factorization, and we develop a confidence-based metric that assists in tracking the convergence of the resonator network.
Authors: Navid Rajabi, Jana Kosecka
Abstract: Vision and Language Models (VLMs) continue to demonstrate remarkable zero-shot (ZS) performance across various tasks. However, many probing studies have revealed that even the best-performing VLMs struggle to capture aspects of compositional scene understanding, lacking the ability to properly ground and localize linguistic phrases in images. Recent VLM advancements include scaling up both model and dataset sizes, additional training objectives and levels of supervision, and variations in the model architectures. To characterize the grounding ability of VLMs, such as phrase grounding, referring expressions comprehension, and relationship understanding, Pointing Game has been used as an evaluation metric for datasets with bounding box annotations. In this paper, we introduce a novel suite of quantitative metrics that utilize GradCAM activations to rigorously evaluate the grounding capabilities of pre-trained VLMs like CLIP, BLIP, and ALBEF. These metrics offer an explainable and quantifiable approach for a more detailed comparison of the zero-shot capabilities of VLMs and enable measuring models' grounding uncertainty. This characterization reveals interesting tradeoffs between the size of the model, the dataset size, and their performance.
Authors: Siyuan Xiang, Chin Tseng, Congcong Wen, Deshana Desai, Yifeng Kou, Binil Starly, Daniele Panozzo, Chen Feng
Abstract: We introduce the first work on benchmarking and evaluating deep clustering algorithms on large-scale non-categorical 3D CAD models. We first propose a workflow to allow expert mechanical engineers to efficiently annotate 252,648 carefully sampled pairwise CAD model similarities, from a subset of the ABC dataset with 22,968 shapes. Using seven baseline deep clustering methods, we then investigate the fundamental challenges of evaluating clustering methods for non-categorical data. Based on these challenges, we propose a novel and viable ensemble-based clustering comparison approach. This work is the first to directly target the underexplored area of deep clustering algorithms for 3D shapes, and we believe it will be an important building block to analyze and utilize the massive 3D shape collections that are starting to appear in deep geometric computing.
Authors: Carlos Eduardo G. R. Alves, Francisco de Assis Boldt, Thiago M. Paix\~ao
Abstract: Effective communication is paramount for the inclusion of deaf individuals in society. However, persistent communication barriers due to limited Sign Language (SL) knowledge hinder their full participation. In this context, Sign Language Recognition (SLR) systems have been developed to improve communication between signing and non-signing individuals. In particular, there is the problem of recognizing isolated signs (Isolated Sign Language Recognition, ISLR) of great relevance in the development of vision-based SL search engines, learning tools, and translation systems. This work proposes an ISLR approach where body, hands, and facial landmarks are extracted throughout time and encoded as 2-D images. These images are processed by a convolutional neural network, which maps the visual-temporal information into a sign label. Experimental results demonstrate that our method surpassed the state-of-the-art in terms of performance metrics on two widely recognized datasets in Brazilian Sign Language (LIBRAS), the primary focus of this study. In addition to being more accurate, our method is more time-efficient and easier to train due to its reliance on a simpler network architecture and solely RGB data as input.
Authors: Evangelos Ververas, Rolandos Alexandros Potamias, Jifei Song, Jiankang Deng, Stefanos Zafeiriou
Abstract: Following the advent of NeRFs, 3D Gaussian Splatting (3D-GS) has paved the way to real-time neural rendering overcoming the computational burden of volumetric methods. Following the pioneering work of 3D-GS, several methods have attempted to achieve compressible and high-fidelity performance alternatives. However, by employing a geometry-agnostic optimization scheme, these methods neglect the inherent 3D structure of the scene, thereby restricting the expressivity and the quality of the representation, resulting in various floating points and artifacts. In this work, we propose a structure-aware Gaussian Splatting method (SAGS) that implicitly encodes the geometry of the scene, which reflects to state-of-the-art rendering performance and reduced storage requirements on benchmark novel-view synthesis datasets. SAGS is founded on a local-global graph representation that facilitates the learning of complex scenes and enforces meaningful point displacements that preserve the scene's geometry. Additionally, we introduce a lightweight version of SAGS, using a simple yet effective mid-point interpolation scheme, which showcases a compact representation of the scene with up to 24$\times$ size reduction without the reliance on any compression strategies. Extensive experiments across multiple benchmark datasets demonstrate the superiority of SAGS compared to state-of-the-art 3D-GS methods under both rendering quality and model size. Besides, we demonstrate that our structure-aware method can effectively mitigate floating artifacts and irregular distortions of previous methods while obtaining precise depth maps. Project page https://eververas.github.io/SAGS/.
Authors: Dongyun Lin, Yi Cheng, Shangbo Mao, Aiyuan Guo, Yiqun Li
Abstract: Large vision-language models have impressively promote the performance of 2D visual recognition under zero/few-shot scenarios. In this paper, we focus on exploiting the large vision-language model, i.e., CLIP, to address zero/few-shot 3D shape recognition based on multi-view representations. The key challenge for both tasks is to generate a discriminative descriptor of the 3D shape represented by multiple view images under the scenarios of either without explicit training (zero-shot 3D shape recognition) or training with a limited number of data (few-shot 3D shape recognition). We analyze that both tasks are relevant and can be considered simultaneously. Specifically, leveraging the descriptor which is effective for zero-shot inference to guide the tuning of the aggregated descriptor under the few-shot training can significantly improve the few-shot learning efficacy. Hence, we propose Prompt-Enhanced View Aggregation Network (PEVA-Net) to simultaneously address zero/few-shot 3D shape recognition. Under the zero-shot scenario, we propose to leverage the prompts built up from candidate categories to enhance the aggregation process of multiple view-associated visual features. The resulting aggregated feature serves for effective zero-shot recognition of the 3D shapes. Under the few-shot scenario, we first exploit a transformer encoder to aggregate the view-associated visual features into a global descriptor. To tune the encoder, together with the main classification loss, we propose a self-distillation scheme via a feature distillation loss by treating the zero-shot descriptor as the guidance signal for the few-shot descriptor. This scheme can significantly enhance the few-shot learning efficacy.
Authors: Cai Yu, Shan Jia, Xiaomeng Fu, Jin Liu, Jiahe Tian, Jiao Dai, Xi Wang, Siwei Lyu, Jizhong Han
Abstract: With the rising prevalence of deepfakes, there is a growing interest in developing generalizable detection methods for various types of deepfakes. While effective in their specific modalities, traditional detection methods fall short in addressing the generalizability of detection across diverse cross-modal deepfakes. This paper aims to explicitly learn potential cross-modal correlation to enhance deepfake detection towards various generation scenarios. Our approach introduces a correlation distillation task, which models the inherent cross-modal correlation based on content information. This strategy helps to prevent the model from overfitting merely to audio-visual synchronization. Additionally, we present the Cross-Modal Deepfake Dataset (CMDFD), a comprehensive dataset with four generation methods to evaluate the detection of diverse cross-modal deepfakes. The experimental results on CMDFD and FakeAVCeleb datasets demonstrate the superior generalizability of our method over existing state-of-the-art methods. Our code and data can be found at \url{https://github.com/ljj898/CMDFD-Dataset-and-Deepfake-Detection}.
URLs: https://github.com/ljj898/CMDFD-Dataset-and-Deepfake-Detection
Authors: Guilherme Potje, Felipe Cadar, Andre Araujo, Renato Martins, Erickson R. Nascimento
Abstract: We introduce a lightweight and accurate architecture for resource-efficient visual correspondence. Our method, dubbed XFeat (Accelerated Features), revisits fundamental design choices in convolutional neural networks for detecting, extracting, and matching local features. Our new model satisfies a critical need for fast and robust algorithms suitable to resource-limited devices. In particular, accurate image matching requires sufficiently large image resolutions - for this reason, we keep the resolution as large as possible while limiting the number of channels in the network. Besides, our model is designed to offer the choice of matching at the sparse or semi-dense levels, each of which may be more suitable for different downstream applications, such as visual navigation and augmented reality. Our model is the first to offer semi-dense matching efficiently, leveraging a novel match refinement module that relies on coarse local descriptors. XFeat is versatile and hardware-independent, surpassing current deep learning-based local features in speed (up to 5x faster) with comparable or better accuracy, proven in pose estimation and visual localization. We showcase it running in real-time on an inexpensive laptop CPU without specialized hardware optimizations. Code and weights are available at www.verlab.dcc.ufmg.br/descriptors/xfeat_cvpr24.
Authors: Benet Oriol Sabat, Alessandro Achille, Matthew Trager, Stefano Soatto
Abstract: We propose NeRF-Insert, a NeRF editing framework that allows users to make high-quality local edits with a flexible level of control. Unlike previous work that relied on image-to-image models, we cast scene editing as an in-painting problem, which encourages the global structure of the scene to be preserved. Moreover, while most existing methods use only textual prompts to condition edits, our framework accepts a combination of inputs of different modalities as reference. More precisely, a user may provide a combination of textual and visual inputs including images, CAD models, and binary image masks for specifying a 3D region. We use generic image generation models to in-paint the scene from multiple viewpoints, and lift the local edits to a 3D-consistent NeRF edit. Compared to previous methods, our results show better visual quality and also maintain stronger consistency with the original NeRF.
Authors: Yoonsik Kim, Moonbin Yim, Ka Yeon Song
Abstract: In this paper, we establish a benchmark for table visual question answering, referred to as the TableVQA-Bench, derived from pre-existing table question-answering (QA) and table structure recognition datasets. It is important to note that existing datasets have not incorporated images or QA pairs, which are two crucial components of TableVQA. As such, the primary objective of this paper is to obtain these necessary components. Specifically, images are sourced either through the application of a \textit{stylesheet} or by employing the proposed table rendering system. QA pairs are generated by exploiting the large language model (LLM) where the input is a text-formatted table. Ultimately, the completed TableVQA-Bench comprises 1,500 QA pairs. We comprehensively compare the performance of various multi-modal large language models (MLLMs) on TableVQA-Bench. GPT-4V achieves the highest accuracy among commercial and open-sourced MLLMs from our experiments. Moreover, we discover that the number of vision queries plays a significant role in TableVQA performance. To further analyze the capabilities of MLLMs in comparison to their LLM backbones, we investigate by presenting image-formatted tables to MLLMs and text-formatted tables to LLMs, respectively. Our findings suggest that processing visual inputs is more challenging than text inputs, as evidenced by the lower performance of MLLMs, despite generally requiring higher computational costs than LLMs. The proposed TableVQA-Bench and evaluation codes are available at \href{https://github.com/naver-ai/tablevqabench}{https://github.com/naver-ai/tablevqabench}.
URLs: https://github.com/naver-ai/tablevqabench, https://github.com/naver-ai/tablevqabench
Authors: Jiading Fang, Xiangshan Tan, Shengjie Lin, Igor Vasiljevic, Vitor Guizilini, Hongyuan Mei, Rares Ambrus, Gregory Shakhnarovich, Matthew R Walter
Abstract: If robots are to work effectively alongside people, they must be able to interpret natural language references to objects in their 3D environment. Understanding 3D referring expressions is challenging -- it requires the ability to both parse the 3D structure of the scene and correctly ground free-form language in the presence of distraction and clutter. We introduce Transcrib3D, an approach that brings together 3D detection methods and the emergent reasoning capabilities of large language models (LLMs). Transcrib3D uses text as the unifying medium, which allows us to sidestep the need to learn shared representations connecting multi-modal inputs, which would require massive amounts of annotated 3D data. As a demonstration of its effectiveness, Transcrib3D achieves state-of-the-art results on 3D reference resolution benchmarks, with a great leap in performance from previous multi-modality baselines. To improve upon zero-shot performance and facilitate local deployment on edge computers and robots, we propose self-correction for fine-tuning that trains smaller models, resulting in performance close to that of large models. We show that our method enables a real robot to perform pick-and-place tasks given queries that contain challenging referring expressions. Project site is at https://ripl.github.io/Transcrib3D.
Authors: Anudeep Das, Vasisht Duddu, Rui Zhang, N. Asokan
Abstract: Diffusion-based text-to-image (T2I) models generate high-fidelity images for given textual prompts. They are trained on large datasets scraped from the Internet, potentially containing unacceptable concepts (e.g., copyright infringing or unsafe). Retraining T2I models after filtering out unacceptable concepts in the training data is inefficient and degrades utility. Hence, there is a need for concept removal techniques (CRTs) which are effective in removing unacceptable concepts, utility-preserving on acceptable concepts, and robust against evasion with adversarial prompts. None of the prior filtering and fine-tuning CRTs satisfy all these requirements simultaneously. We introduce Espresso, the first robust concept filter based on Contrastive Language-Image Pre-Training (CLIP). It identifies unacceptable concepts by projecting the generated image's embedding onto the vector connecting unacceptable and acceptable concepts in the joint text-image embedding space. This ensures robustness by restricting the adversary to adding noise only along this vector, in the direction of the acceptable concept. Further fine-tuning Espresso to separate embeddings of acceptable and unacceptable concepts, while preserving their pairing with image embeddings, ensures both effectiveness and utility. We evaluate Espresso on eleven concepts to show that it is effective (~5% CLIP accuracy on unacceptable concepts), utility-preserving (~93% normalized CLIP score on acceptable concepts), and robust (~4% CLIP accuracy on adversarial prompts for unacceptable concepts). Finally, we present theoretical bounds for the certified robustness of Espresso against adversarial prompts, and an empirical analysis.
Authors: Xin Ma, Puchen Zhu, Xiao Li, Xiaoyin Zheng, Jianshu Zhou, Xuchen Wang, Kwok Wai Samuel Au
Abstract: Depth position highly affects lens distortion, especially in close-range photography, which limits the measurement accuracy of existing stereo vision systems. Moreover, traditional depth-dependent distortion models and their calibration methods have remained complicated. In this work, we propose a minimal set of parameters based depth-dependent distortion model (MDM), which considers the radial and decentering distortions of the lens to improve the accuracy of stereo vision systems and simplify their calibration process. In addition, we present an easy and flexible calibration method for the MDM of stereo vision systems with a commonly used planar pattern, which requires cameras to observe the planar pattern in different orientations. The proposed technique is easy to use and flexible compared with classical calibration techniques for depth-dependent distortion models in which the lens must be perpendicular to the planar pattern. The experimental validation of the MDM and its calibration method showed that the MDM improved the calibration accuracy by 56.55% and 74.15% compared with the Li's distortion model and traditional Brown's distortion model. Besides, an iteration-based reconstruction method is proposed to iteratively estimate the depth information in the MDM during three-dimensional reconstruction. The results showed that the accuracy of the iteration-based reconstruction method was improved by 9.08% compared with that of the non-iteration reconstruction method.
Authors: Junghyup lee, Dohyung Kim, Jeimin Jeon, Bumsub Ham
Abstract: Quantization-aware training (QAT) simulates a quantization process during training to lower bit-precision of weights/activations. It learns quantized weights indirectly by updating latent weights, i.e., full-precision inputs to a quantizer, using gradient-based optimizers. We claim that coupling a user-defined learning rate (LR) with these optimizers is sub-optimal for QAT. Quantized weights transit discrete levels of a quantizer, only if corresponding latent weights pass transition points, where the quantizer changes discrete states. This suggests that the changes of quantized weights are affected by both the LR for latent weights and their distributions. It is thus difficult to control the degree of changes for quantized weights by scheduling the LR manually. We conjecture that the degree of parameter changes in QAT is related to the number of quantized weights transiting discrete levels. Based on this, we introduce a transition rate (TR) scheduling technique that controls the number of transitions of quantized weights explicitly. Instead of scheduling a LR for latent weights, we schedule a target TR of quantized weights, and update the latent weights with a novel transition-adaptive LR (TALR), enabling considering the degree of changes for the quantized weights during QAT. Experimental results demonstrate the effectiveness of our approach on standard benchmarks.
Authors: Jeonghoon Park, Chaeyeon Chung, Juyoung Lee, Jaegul Choo
Abstract: In the image classification task, deep neural networks frequently rely on bias attributes that are spuriously correlated with a target class in the presence of dataset bias, resulting in degraded performance when applied to data without bias attributes. The task of debiasing aims to compel classifiers to learn intrinsic attributes that inherently define a target class rather than focusing on bias attributes. While recent approaches mainly focus on emphasizing the learning of data samples without bias attributes (i.e., bias-conflicting samples) compared to samples with bias attributes (i.e., bias-aligned samples), they fall short of directly guiding models where to focus for learning intrinsic features. To address this limitation, this paper proposes a method that provides the model with explicit spatial guidance that indicates the region of intrinsic features. We first identify the intrinsic features by investigating the class-discerning common features between a bias-aligned (BA) sample and a bias-conflicting (BC) sample (i.e., bias-contrastive pair). Next, we enhance the intrinsic features in the BA sample that are relatively under-exploited for prediction compared to the BC sample. To construct the bias-contrastive pair without using bias information, we introduce a bias-negative score that distinguishes BC samples from BA samples employing a biased model. The experiments demonstrate that our method achieves state-of-the-art performance on synthetic and real-world datasets with various levels of bias severity.
Authors: Taylor Archibald, Tony Martinez
Abstract: Document semantic segmentation is a promising avenue that can facilitate document analysis tasks, including optical character recognition (OCR), form classification, and document editing. Although several synthetic datasets have been developed to distinguish handwriting from printed text, they fall short in class variety and document diversity. We demonstrate the limitations of training on existing datasets when solving the National Archives Form Semantic Segmentation dataset (NAFSS), a dataset which we introduce. To address these limitations, we propose the most comprehensive document semantic segmentation synthesis pipeline to date, incorporating preprinted text, handwriting, and document backgrounds from over 10 sources to create the Document Element Layer INtegration Ensemble 8K, or DELINE8K dataset. Our customized dataset exhibits superior performance on the NAFSS benchmark, demonstrating it as a promising tool in further research. The DELINE8K dataset is available at https://github.com/Tahlor/deline8k.
Authors: Zhenglin Li, Bo Guan, Yuanzhou Wei, Yiming Zhou, Jingyu Zhang, Jinxin Xu
Abstract: Generative Adversarial Networks (GANs) have significantly advanced image processing, with Pix2Pix being a notable framework for image-to-image translation. This paper explores a novel application of Pix2Pix to transform abstract map images into realistic ground truth images, addressing the scarcity of such images crucial for domains like urban planning and autonomous vehicle training. We detail the Pix2Pix model's utilization for generating high-fidelity datasets, supported by a dataset of paired map and aerial images, and enhanced by a tailored training regimen. The results demonstrate the model's capability to accurately render complex urban features, establishing its efficacy and potential for broad real-world applications.
Authors: Sairam VC Rebbapragada, Pranoy Panda, Vineeth N Balasubramanian
Abstract: A vision-based drone-to-drone detection system is crucial for various applications like collision avoidance, countering hostile drones, and search-and-rescue operations. However, detecting drones presents unique challenges, including small object sizes, distortion, occlusion, and real-time processing requirements. Current methods integrating multi-scale feature fusion and temporal information have limitations in handling extreme blur and minuscule objects. To address this, we propose a novel coarse-to-fine detection strategy based on vision transformers. We evaluate our approach on three challenging drone-to-drone detection datasets, achieving F1 score enhancements of 7%, 3%, and 1% on the FL-Drones, AOT, and NPS-Drones datasets, respectively. Additionally, we demonstrate real-time processing capabilities by deploying our model on an edge-computing device. Our code will be made publicly available.
Authors: Wentao Lei, Li Liu, Jun Wang
Abstract: Cued Speech (CS) is an advanced visual phonetic encoding system that integrates lip reading with hand codings, enabling people with hearing impairments to communicate efficiently. CS video generation aims to produce specific lip and gesture movements of CS from audio or text inputs. The main challenge is that given limited CS data, we strive to simultaneously generate fine-grained hand and finger movements, as well as lip movements, meanwhile the two kinds of movements need to be asynchronously aligned. Existing CS generation methods are fragile and prone to poor performance due to template-based statistical models and careful hand-crafted pre-processing to fit the models. Therefore, we propose a novel Gloss-prompted Diffusion-based CS Gesture generation framework (called GlossDiff). Specifically, to integrate additional linguistic rules knowledge into the model. we first introduce a bridging instruction called \textbf{Gloss}, which is an automatically generated descriptive text to establish a direct and more delicate semantic connection between spoken language and CS gestures. Moreover, we first suggest rhythm is an important paralinguistic feature for CS to improve the communication efficacy. Therefore, we propose a novel Audio-driven Rhythmic Module (ARM) to learn rhythm that matches audio speech. Moreover, in this work, we design, record, and publish the first Chinese CS dataset with four CS cuers. Extensive experiments demonstrate that our method quantitatively and qualitatively outperforms current state-of-the-art (SOTA) methods. We release the code and data at https://glossdiff.github.io/.
Authors: Xingyu Song, Zhan Li, Shi Chen, Kazuyuki Demachi
Abstract: 3D human pose estimation is a vital task in computer vision, involving the prediction of human joint positions from images or videos to reconstruct a skeleton of a human in three-dimensional space. This technology is pivotal in various fields, including animation, security, human-computer interaction, and automotive safety, where it promotes both technological progress and enhanced human well-being. The advent of deep learning significantly advances the performance of 3D pose estimation by incorporating temporal information for predicting the spatial positions of human joints. However, traditional methods often fall short as they primarily focus on the spatial coordinates of joints and overlook the orientation and rotation of the connecting bones, which are crucial for a comprehensive understanding of human pose in 3D space. To address these limitations, we introduce Quater-GCN (Q-GCN), a directed graph convolutional network tailored to enhance pose estimation by orientation. Q-GCN excels by not only capturing the spatial dependencies among node joints through their coordinates but also integrating the dynamic context of bone rotations in 2D space. This approach enables a more sophisticated representation of human poses by also regressing the orientation of each bone in 3D space, moving beyond mere coordinate prediction. Furthermore, we complement our model with a semi-supervised training strategy that leverages unlabeled data, addressing the challenge of limited orientation ground truth data. Through comprehensive evaluations, Q-GCN has demonstrated outstanding performance against current state-of-the-art methods.
Authors: Shuanghao Bai, Yuedi Zhang, Wanqi Zhou, Zhirong Luan, Badong Chen
Abstract: Large pre-trained vision language models (VLMs) have shown impressive zero-shot ability on downstream tasks with manually designed prompt, which are not optimal for specific domains. To further adapt VLMs to downstream tasks, soft prompt is proposed to replace manually designed prompt, which acts as a learning vector that undergoes fine-tuning based on specific domain data. Prior prompt learning methods primarily learn a fixed prompt and residuled prompt from training samples. However, the learned prompts lack diversity and ignore information about unseen domains, potentially compromising the transferability of the prompts. In this paper, we reframe the prompt learning framework from a generative perspective and propose a simple yet efficient method for the Domain Generalization (DG) task, namely \textbf{S}oft \textbf{P}rompt \textbf{G}eneration (SPG). To the best of our knowledge, we are the first to introduce the generative model into prompt learning in VLMs and explore its potential for producing soft prompts by relying solely on the generative model, ensuring the diversity of prompts. Specifically, SPG consists of a two-stage training phase and an inference phase. During the training phase, we introduce soft prompt labels for each domain, aiming to incorporate the generative model domain knowledge. During the inference phase, the generator of the generative model is employed to obtain instance-specific soft prompts for the unseen target domain. Extensive experiments on five domain generalization benchmarks of three DG tasks demonstrate that our proposed SPG achieves state-of-the-art performance. The code will be available soon.
Authors: Wanqi Zhou, Shuanghao Bai, Qibin Zhao, Badong Chen
Abstract: Pretrained vision-language models (VLMs) like CLIP have shown impressive generalization performance across various downstream tasks, yet they remain vulnerable to adversarial attacks. While prior research has primarily concentrated on improving the adversarial robustness of image encoders to guard against attacks on images, the exploration of text-based and multimodal attacks has largely been overlooked. In this work, we initiate the first known and comprehensive effort to study adapting vision-language models for adversarial robustness under the multimodal attack. Firstly, we introduce a multimodal attack strategy and investigate the impact of different attacks. We then propose a multimodal contrastive adversarial training loss, aligning the clean and adversarial text embeddings with the adversarial and clean visual features, to enhance the adversarial robustness of both image and text encoders of CLIP. Extensive experiments on 15 datasets across two tasks demonstrate that our method significantly improves the adversarial robustness of CLIP. Interestingly, we find that the model fine-tuned against multimodal adversarial attacks exhibits greater robustness than its counterpart fine-tuned solely against image-based attacks, even in the context of image attacks, which may open up new possibilities for enhancing the security of VLMs.
Authors: Yun-Hao Cao, Jianxin Wu
Abstract: Self-supervised learning (SSL) has developed rapidly in recent years. However, most of the mainstream methods are computationally expensive and rely on two (or more) augmentations for each image to construct positive pairs. Moreover, they mainly focus on large models and large-scale datasets, which lack flexibility and feasibility in many practical applications. In this paper, we propose an efficient single-branch SSL method based on non-parametric instance discrimination, aiming to improve the algorithm, model, and data efficiency of SSL. By analyzing the gradient formula, we correct the update rule of the memory bank with improved performance. We further propose a novel self-distillation loss that minimizes the KL divergence between the probability distribution and its square root version. We show that this alleviates the infrequent updating problem in instance discrimination and greatly accelerates convergence. We systematically compare the training overhead and performance of different methods in different scales of data, and under different backbones. Experimental results show that our method outperforms various baselines with significantly less overhead, and is especially effective for limited amounts of data and small models.
Authors: Jinyoung Jun, Jae-Han Lee, Chang-Su Kim
Abstract: The main function of depth completion is to compensate for an insufficient and unpredictable number of sparse depth measurements of hardware sensors. However, existing research on depth completion assumes that the sparsity -- the number of points or LiDAR lines -- is fixed for training and testing. Hence, the completion performance drops severely when the number of sparse depths changes significantly. To address this issue, we propose the sparsity-adaptive depth refinement (SDR) framework, which refines monocular depth estimates using sparse depth points. For SDR, we propose the masked spatial propagation network (MSPN) to perform SDR with a varying number of sparse depths effectively by gradually propagating sparse depth information throughout the entire depth map. Experimental results demonstrate that MPSN achieves state-of-the-art performance on both SDR and conventional depth completion scenarios.
Authors: Sungjune Park, Hyunjun Kim, Yong Man Ro
Abstract: Pedestrian detection is a crucial field of computer vision research which can be adopted in various real-world applications (e.g., self-driving systems). However, despite noticeable evolution of pedestrian detection, pedestrian representations learned within a detection framework are usually limited to particular scene data in which they were trained. Therefore, in this paper, we propose a novel approach to construct versatile pedestrian knowledge bank containing representative pedestrian knowledge which can be applicable to various detection frameworks and adopted in diverse scenes. We extract generalized pedestrian knowledge from a large-scale pretrained model, and we curate them by quantizing most representative features and guiding them to be distinguishable from background scenes. Finally, we construct versatile pedestrian knowledge bank which is composed of such representations, and then we leverage it to complement and enhance pedestrian features within a pedestrian detection framework. Through comprehensive experiments, we validate the effectiveness of our method, demonstrating its versatility and outperforming state-of-the-art detection performances.
Authors: Wang Zhang, Tingting Li, Yuntian Zhang, Gensheng Pei, Xiruo Jiang, Yazhou Yao
Abstract: Matching visible and near-infrared (NIR) images remains a significant challenge in remote sensing image fusion. The nonlinear radiometric differences between heterogeneous remote sensing images make the image matching task even more difficult. Deep learning has gained substantial attention in computer vision tasks in recent years. However, many methods rely on supervised learning and necessitate large amounts of annotated data. Nevertheless, annotated data is frequently limited in the field of remote sensing image matching. To address this challenge, this paper proposes a novel keypoint descriptor approach that obtains robust feature descriptors via a self-supervised matching network. A light-weight transformer network, termed as LTFormer, is designed to generate deep-level feature descriptors. Furthermore, we implement an innovative triplet loss function, LT Loss, to enhance the matching performance further. Our approach outperforms conventional hand-crafted local feature descriptors and proves equally competitive compared to state-of-the-art deep learning-based methods, even amidst the shortage of annotated data.
Authors: Sol\`ene Tarride, Christopher Kermorvant
Abstract: In recent advances in automatic text recognition (ATR), deep neural networks have demonstrated the ability to implicitly capture language statistics, potentially reducing the need for traditional language models. This study directly addresses whether explicit language models, specifically n-gram models, still contribute to the performance of state-of-the-art deep learning architectures in the field of handwriting recognition. We evaluate two prominent neural network architectures, PyLaia and DAN, with and without the integration of explicit n-gram language models. Our experiments on three datasets - IAM, RIMES, and NorHand v2 - at both line and page level, investigate optimal parameters for n-gram models, including their order, weight, smoothing methods and tokenization level. The results show that incorporating character or subword n-gram models significantly improves the performance of ATR models on all datasets, challenging the notion that deep learning models alone are sufficient for optimal performance. In particular, the combination of DAN with a character language model outperforms current benchmarks, confirming the value of hybrid approaches in modern document analysis systems.
Authors: Lingyi Hong, Zhongying Liu, Wenchao Chen, Chenzhi Tan, Yuang Feng, Xinyu Zhou, Pinxue Guo, Jinglun Li, Zhaoyu Chen, Shuyong Gao, Wei Zhang, Wenqiang Zhang
Abstract: Video object segmentation (VOS) aims to distinguish and track target objects in a video. Despite the excellent performance achieved by off-the-shell VOS models, existing VOS benchmarks mainly focus on short-term videos lasting about 5 seconds, where objects remain visible most of the time. However, these benchmarks poorly represent practical applications, and the absence of long-term datasets restricts further investigation of VOS in realistic scenarios. Thus, we propose a novel benchmark named LVOS, comprising 720 videos with 296,401 frames and 407,945 high-quality annotations. Videos in LVOS last 1.14 minutes on average, approximately 5 times longer than videos in existing datasets. Each video includes various attributes, especially challenges deriving from the wild, such as long-term reappearing and cross-temporal similar objects. Compared to previous benchmarks, our LVOS better reflects VOS models' performance in real scenarios. Based on LVOS, we evaluate 20 existing VOS models under 4 different settings and conduct a comprehensive analysis. On LVOS, these models suffer a large performance drop, highlighting the challenge of achieving precise tracking and segmentation in real-world scenarios. Attribute-based analysis indicates that key factor to accuracy decline is the increased video length, emphasizing LVOS's crucial role. We hope our LVOS can advance development of VOS in real scenes. Data and code are available at https://lingyihongfd.github.io/lvos.github.io/.
Authors: Thomas Constum, Lucas Preel, Th\'eo Larcher, Pierrick Tranouez, Thierry Paquet, Sandra Br\'ee
Abstract: The EXO-POPP project aims to establish a comprehensive database comprising 300,000 marriage records from Paris and its suburbs, spanning the years 1880 to 1940, which are preserved in over 130,000 scans of double pages. Each marriage record may encompass up to 118 distinct types of information that require extraction from plain text. In this paper, we introduce the M-POPP dataset, a subset of the M-POPP database with annotations for full-page text recognition and information extraction in both handwritten and printed documents, and which is now publicly available. We present a fully end-to-end architecture adapted from the DAN, designed to perform both handwritten text recognition and information extraction directly from page images without the need for explicit segmentation. We showcase the information extraction capabilities of this architecture by achieving a new state of the art for full-page Information Extraction on Esposalles and we use this architecture as a baseline for the M-POPP dataset. We also assess and compare how different encoding strategies for named entities in the text affect the performance of jointly recognizing handwritten text and extracting information, from full pages.
Authors: Zhanwei Zhang, Zishuo Hua, Minghao Chen, Wei Lu, Binbin Lin, Deng Cai, Wenxiao Wang
Abstract: Predicting future trajectories of traffic agents accurately holds substantial importance in various applications such as autonomous driving. Previous methods commonly infer all future steps of an agent either recursively or simultaneously. However, the recursive strategy suffers from the accumulated error, while the simultaneous strategy overlooks the constraints among future steps, resulting in kinematically infeasible predictions. To address these issues, in this paper, we propose G2LTraj, a plug-and-play global-to-local generation approach for trajectory prediction. Specifically, we generate a series of global key steps that uniformly cover the entire future time range. Subsequently, the local intermediate steps between the adjacent key steps are recursively filled in. In this way, we prevent the accumulated error from propagating beyond the adjacent key steps. Moreover, to boost the kinematical feasibility, we not only introduce the spatial constraints among key steps but also strengthen the temporal constraints among the intermediate steps. Finally, to ensure the optimal granularity of key steps, we design a selectable granularity strategy that caters to each predicted trajectory. Our G2LTraj significantly improves the performance of seven existing trajectory predictors across the ETH, UCY and nuScenes datasets. Experimental results demonstrate its effectiveness. Code will be available at https://github.com/Zhanwei-Z/G2LTraj.
Authors: Minghao Han, Xukun Zhang, Dingkang Yang, Tao Liu, Haopeng Kuang, Jinghui Feng, Lihua Zhang
Abstract: Survival prediction is a complex ordinal regression task that aims to predict the survival coefficient ranking among a cohort of patients, typically achieved by analyzing patients' whole slide images. Existing deep learning approaches mainly adopt multiple instance learning or graph neural networks under weak supervision. Most of them are unable to uncover the diverse interactions between different types of biological entities(\textit{e.g.}, cell cluster and tissue block) across multiple scales, while such interactions are crucial for patient survival prediction. In light of this, we propose a novel multi-scale heterogeneity-aware hypergraph representation framework. Specifically, our framework first constructs a multi-scale heterogeneity-aware hypergraph and assigns each node with its biological entity type. It then mines diverse interactions between nodes on the graph structure to obtain a global representation. Experimental results demonstrate that our method outperforms state-of-the-art approaches on three benchmark datasets. Code is publicly available at \href{https://github.com/Hanminghao/H2GT}{https://github.com/Hanminghao/H2GT}.
URLs: https://github.com/Hanminghao/H2GT, https://github.com/Hanminghao/H2GT
Authors: Soham Mitra, Atri Sukul, Swalpa Kumar Roy, Pravendra Singh, Vinay Verma
Abstract: Deep learning models have achieved remarkable success across diverse domains. However, the intricate nature of these models often impedes a clear understanding of their decision-making processes. This is where Explainable AI (XAI) becomes indispensable, offering intuitive explanations for model decisions. In this work, we propose a simple yet highly effective approach, ScoreCAM++, which introduces modifications to enhance the promising ScoreCAM method for visual explainability. Our proposed approach involves altering the normalization function within the activation layer utilized in ScoreCAM, resulting in significantly improved results compared to previous efforts. Additionally, we apply an activation function to the upsampled activation layers to enhance interpretability. This improvement is achieved by selectively gating lower-priority values within the activation layer. Through extensive experiments and qualitative comparisons, we demonstrate that ScoreCAM++ consistently achieves notably superior performance and fairness in interpreting the decision-making process compared to both ScoreCAM and previous methods.
Authors: Hao-Cheng Lo, Jung-Mei Chu, Jieh Hsiang, Chun-Chieh Cho
Abstract: In patent prosecution, image-based retrieval systems for identifying similarities between current patent images and prior art are pivotal to ensure the novelty and non-obviousness of patent applications. Despite their growing popularity in recent years, existing attempts, while effective at recognizing images within the same patent, fail to deliver practical value due to their limited generalizability in retrieving relevant prior art. Moreover, this task inherently involves the challenges posed by the abstract visual features of patent images, the skewed distribution of image classifications, and the semantic information of image descriptions. Therefore, we propose a language-informed, distribution-aware multimodal approach to patent image feature learning, which enriches the semantic understanding of patent image by integrating Large Language Models and improves the performance of underrepresented classes with our proposed distribution-aware contrastive losses. Extensive experiments on DeepPatent2 dataset show that our proposed method achieves state-of-the-art or comparable performance in image-based patent retrieval with mAP +53.3%, Recall@10 +41.8%, and MRR@10 +51.9%. Furthermore, through an in-depth user analysis, we explore our model in aiding patent professionals in their image retrieval efforts, highlighting the model's real-world applicability and effectiveness.
Authors: Zhigang Sun, Zixu Wang, Lavdim Halilaj, Juergen Luettin
Abstract: Trajectory prediction in autonomous driving relies on accurate representation of all relevant contexts of the driving scene including traffic participants, road topology, traffic signs as well as their semantic relations to each other. Despite increased attention to this issue, most approaches in trajectory prediction do not consider all of these factors sufficiently. This paper describes a method SemanticFormer to predict multimodal trajectories by reasoning over a semantic traffic scene graph using a hybrid approach. We extract high-level information in the form of semantic meta-paths from a knowledge graph which is then processed by a novel pipeline based on multiple attention mechanisms to predict accurate trajectories. The proposed architecture comprises a hierarchical heterogeneous graph encoder, which can capture spatio-temporal and relational information across agents and between agents and road elements, and a predictor that fuses the different encodings and decodes trajectories with probabilities. Finally, a refinement module evaluates permitted meta-paths of trajectories and speed profiles to obtain final predicted trajectories. Evaluation of the nuScenes benchmark demonstrates improved performance compared to the state-of-the-art methods.
Authors: Xiaoxuan Han, Songlin Yang, Wei Wang, Yang Li, Jing Dong
Abstract: Advanced text-to-image diffusion models raise safety concerns regarding identity privacy violation, copyright infringement, and Not Safe For Work content generation. Towards this, unlearning methods have been developed to erase these involved concepts from diffusion models. However, these unlearning methods only shift the text-to-image mapping and preserve the visual content within the generative space of diffusion models, leaving a fatal flaw for restoring these erased concepts. This erasure trustworthiness problem needs probe, but previous methods are sub-optimal from two perspectives: (1) Lack of transferability: Some methods operate within a white-box setting, requiring access to the unlearned model. And the learned adversarial input often fails to transfer to other unlearned models for concept restoration; (2) Limited attack: The prompt-level methods struggle to restore narrow concepts from unlearned models, such as celebrity identity. Therefore, this paper aims to leverage the transferability of the adversarial attack to probe the unlearning robustness under a black-box setting. This challenging scenario assumes that the unlearning method is unknown and the unlearned model is inaccessible for optimization, requiring the attack to be capable of transferring across different unlearned models. Specifically, we employ an adversarial search strategy to search for the adversarial embedding which can transfer across different unlearned models. This strategy adopts the original Stable Diffusion model as a surrogate model to iteratively erase and search for embeddings, enabling it to find the embedding that can restore the target concept for different unlearning methods. Extensive experiments demonstrate the transferability of the searched adversarial embedding across several state-of-the-art unlearning methods and its effectiveness for different levels of concepts.
Authors: Zhendong Liu, Haifeng Xia, Tong Guo, Libo Sun, Ming Shao, Siyu Xia
Abstract: Human action video recognition has recently attracted more attention in applications such as video security and sports posture correction. Popular solutions, including graph convolutional networks (GCNs) that model the human skeleton as a spatiotemporal graph, have proven very effective. GCNs-based methods with stacked blocks usually utilize top-layer semantics for classification/annotation purposes. Although the global features learned through the procedure are suitable for the general classification, they have difficulty capturing fine-grained action change across adjacent frames -- decisive factors in sports actions. In this paper, we propose a novel ``Cross-block Fine-grained Semantic Cascade (CFSC)'' module to overcome this challenge. In summary, the proposed CFSC progressively integrates shallow visual knowledge into high-level blocks to allow networks to focus on action details. In particular, the CFSC module utilizes the GCN feature maps produced at different levels, as well as aggregated features from proceeding levels to consolidate fine-grained features. In addition, a dedicated temporal convolution is applied at each level to learn short-term temporal features, which will be carried over from shallow to deep layers to maximize the leverage of low-level details. This cross-block feature aggregation methodology, capable of mitigating the loss of fine-grained information, has resulted in improved performance. Last, FD-7, a new action recognition dataset for fencing sports, was collected and will be made publicly available. Experimental results and empirical analysis on public benchmarks (FSD-10) and self-collected (FD-7) demonstrate the advantage of our CFSC module on learning discriminative patterns for action classification over others.
Authors: Zhanwei Zhang, Minghao Chen, Shuai Xiao, Liang Peng, Hengjia Li, Binbin Lin, Ping Li, Wenxiao Wang, Boxi Wu, Deng Cai
Abstract: Recent self-training techniques have shown notable improvements in unsupervised domain adaptation for 3D object detection (3D UDA). These techniques typically select pseudo labels, i.e., 3D boxes, to supervise models for the target domain. However, this selection process inevitably introduces unreliable 3D boxes, in which 3D points cannot be definitively assigned as foreground or background. Previous techniques mitigate this by reweighting these boxes as pseudo labels, but these boxes can still poison the training process. To resolve this problem, in this paper, we propose a novel pseudo label refinery framework. Specifically, in the selection process, to improve the reliability of pseudo boxes, we propose a complementary augmentation strategy. This strategy involves either removing all points within an unreliable box or replacing it with a high-confidence box. Moreover, the point numbers of instances in high-beam datasets are considerably higher than those in low-beam datasets, also degrading the quality of pseudo labels during the training process. We alleviate this issue by generating additional proposals and aligning RoI features across different domains. Experimental results demonstrate that our method effectively enhances the quality of pseudo labels and consistently surpasses the state-of-the-art methods on six autonomous driving benchmarks. Code will be available at https://github.com/Zhanwei-Z/PERE.
Authors: Weiquan Huang, Yifei Shen, Yifan Yang
Abstract: State space models and Mamba-based models have been increasingly applied across various domains, achieving state-of-the-art performance. This technical report introduces the first attempt to train a transferable Mamba model utilizing contrastive language-image pretraining (CLIP). We have trained Mamba models of varying sizes and undertaken comprehensive evaluations of these models on 26 zero-shot classification datasets and 16 out-of-distribution (OOD) datasets. Our findings reveal that a Mamba model with 67 million parameters is on par with a 307 million-parameter Vision Transformer (ViT) model in zero-shot classification tasks, highlighting the parameter efficiency of Mamba models. In tests of OOD generalization, Mamba-based models exhibit exceptional performance in conditions of OOD image contrast or when subjected to high-pass filtering. However, a Hessian analysis indicates that Mamba models feature a sharper and more non-convex landscape compared to ViT-based models, making them more challenging to train. The source code is available at https://github.com/raytrun/mamba-clip.
Authors: Sheng Jin, Ruijie Yao, Lumin Xu, Wentao Liu, Chen Qian, Ji Wu, Ping Luo
Abstract: Instance perception tasks (object detection, instance segmentation, pose estimation, counting) play a key role in industrial applications of visual models. As supervised learning methods suffer from high labeling cost, few-shot learning methods which effectively learn from a limited number of labeled examples are desired. Existing few-shot learning methods primarily focus on a restricted set of tasks, presumably due to the challenges involved in designing a generic model capable of representing diverse tasks in a unified manner. In this paper, we propose UniFS, a universal few-shot instance perception model that unifies a wide range of instance perception tasks by reformulating them into a dynamic point representation learning framework. Additionally, we propose Structure-Aware Point Learning (SAPL) to exploit the higher-order structural relationship among points to further enhance representation learning. Our approach makes minimal assumptions about the tasks, yet it achieves competitive results compared to highly specialized and well optimized specialist models. Codes will be released soon.
Authors: Wen Yin, Jian Lou, Pan Zhou, Yulai Xie, Dan Feng, Yuhua Sun, Tailai Zhang, Lichao Sun
Abstract: Backdoor attacks have been well-studied in visible light object detection (VLOD) in recent years. However, VLOD can not effectively work in dark and temperature-sensitive scenarios. Instead, thermal infrared object detection (TIOD) is the most accessible and practical in such environments. In this paper, our team is the first to investigate the security vulnerabilities associated with TIOD in the context of backdoor attacks, spanning both the digital and physical realms. We introduce two novel types of backdoor attacks on TIOD, each offering unique capabilities: Object-affecting Attack and Range-affecting Attack. We conduct a comprehensive analysis of key factors influencing trigger design, which include temperature, size, material, and concealment. These factors, especially temperature, significantly impact the efficacy of backdoor attacks on TIOD. A thorough understanding of these factors will serve as a foundation for designing physical triggers and temperature controlling experiments. Our study includes extensive experiments conducted in both digital and physical environments. In the digital realm, we evaluate our approach using benchmark datasets for TIOD, achieving an Attack Success Rate (ASR) of up to 98.21%. In the physical realm, we test our approach in two real-world settings: a traffic intersection and a parking lot, using a thermal infrared camera. Here, we attain an ASR of up to 98.38%.
Authors: Chanran Kim, Jeongin Lee, Shichang Joung, Bongmo Kim, Yeul-Min Baek
Abstract: In the field of personalized image generation, the ability to create images preserving concepts has significantly improved. Creating an image that naturally integrates multiple concepts in a cohesive and visually appealing composition can indeed be challenging. This paper introduces "InstantFamily," an approach that employs a novel masked cross-attention mechanism and a multimodal embedding stack to achieve zero-shot multi-ID image generation. Our method effectively preserves ID as it utilizes global and local features from a pre-trained face recognition model integrated with text conditions. Additionally, our masked cross-attention mechanism enables the precise control of multi-ID and composition in the generated images. We demonstrate the effectiveness of InstantFamily through experiments showing its dominance in generating images with multi-ID, while resolving well-known multi-ID generation problems. Additionally, our model achieves state-of-the-art performance in both single-ID and multi-ID preservation. Furthermore, our model exhibits remarkable scalability with a greater number of ID preservation than it was originally trained with.
Authors: Jie Hu, Yawen Huang, Yilin Lu, Guoyang Xie, Guannan Jiang, Yefeng Zheng
Abstract: Anomaly synthesis is one of the effective methods to augment abnormal samples for training. However, current anomaly synthesis methods predominantly rely on texture information as input, which limits the fidelity of synthesized abnormal samples. Because texture information is insufficient to correctly depict the pattern of anomalies, especially for logical anomalies. To surmount this obstacle, we present the AnomalyXFusion framework, designed to harness multi-modality information to enhance the quality of synthesized abnormal samples. The AnomalyXFusion framework comprises two distinct yet synergistic modules: the Multi-modal In-Fusion (MIF) module and the Dynamic Dif-Fusion (DDF) module. The MIF module refines modality alignment by aggregating and integrating various modality features into a unified embedding space, termed X-embedding, which includes image, text, and mask features. Concurrently, the DDF module facilitates controlled generation through an adaptive adjustment of X-embedding conditioned on the diffusion steps. In addition, to reveal the multi-modality representational power of AnomalyXFusion, we propose a new dataset, called MVTec Caption. More precisely, MVTec Caption extends 2.2k accurate image-mask-text annotations for the MVTec AD and LOCO datasets. Comprehensive evaluations demonstrate the effectiveness of AnomalyXFusion, especially regarding the fidelity and diversity for logical anomalies. Project page: http:github.com/hujiecpp/MVTec-Caption
Authors: Teng Zhou, Yongchuan Tang
Abstract: Diffusion models have emerged as effective tools for generating diverse and high-quality content. However, their capability in high-resolution image generation, particularly for panoramic images, still faces challenges such as visible seams and incoherent transitions. In this paper, we propose TwinDiffusion, an optimized framework designed to address these challenges through two key innovations: Crop Fusion for quality enhancement and Cross Sampling for efficiency optimization. We introduce a training-free optimizing stage to refine the similarity of the adjacent image areas, as well as an interleaving sampling strategy to yield dynamic patches during the cropping process. A comprehensive evaluation is conducted to compare TwinDiffusion with the existing methods, considering factors including coherence, fidelity, compatibility, and efficiency. The results demonstrate the superior performance of our approach in generating seamless and coherent panoramas, setting a new standard in quality and efficiency for panoramic image generation.
Authors: Yufeng Yang, Adrian Kneip, Charlotte Frenkel
Abstract: Edge vision systems combining sensing and embedded processing promise low-latency, decentralized, and energy-efficient solutions that forgo reliance on the cloud. As opposed to conventional frame-based vision sensors, event-based cameras deliver a microsecond-scale temporal resolution with sparse information encoding, thereby outlining new opportunities for edge vision systems. However, mainstream algorithms for frame-based vision, which mostly rely on convolutional neural networks (CNNs), can hardly exploit the advantages of event-based vision as they are typically optimized for dense matrix-vector multiplications. While event-driven graph neural networks (GNNs) have recently emerged as a promising solution for sparse event-based vision, their irregular structure is a challenge that currently hinders the design of efficient hardware accelerators. In this paper, we propose EvGNN, the first event-driven GNN accelerator for low-footprint, ultra-low-latency, and high-accuracy edge vision with event-based cameras. It relies on three central ideas: (i) directed dynamic graphs exploiting single-hop nodes with edge-free storage, (ii) event queues for the efficient identification of local neighbors within a spatiotemporally decoupled search range, and (iii) a novel layer-parallel processing scheme enabling the low-latency execution of multi-layer GNNs. We deployed EvGNN on a Xilinx KV260 Ultrascale+ MPSoC platform and benchmarked it on the N-CARS dataset for car recognition, demonstrating a classification accuracy of 87.8% and an average latency per event of 16$\mu$s, thereby enabling real-time, microsecond-resolution event-based vision at the edge.
Authors: Ziyan Chen, Jingwen He, Xinqi Lin, Yu Qiao, Chao Dong
Abstract: Blind face restoration (BFR) on images has significantly progressed over the last several years, while real-world video face restoration (VFR), which is more challenging for more complex face motions such as moving gaze directions and facial orientations involved, remains unsolved. Typical BFR methods are evaluated on privately synthesized datasets or self-collected real-world low-quality face images, which are limited in their coverage of real-world video frames. In this work, we introduced new real-world datasets named FOS with a taxonomy of "Full, Occluded, and Side" faces from mainly video frames to study the applicability of current methods on videos. Compared with existing test datasets, FOS datasets cover more diverse degradations and involve face samples from more complex scenarios, which helps to revisit current face restoration approaches more comprehensively. Given the established datasets, we benchmarked both the state-of-the-art BFR methods and the video super resolution (VSR) methods to comprehensively study current approaches, identifying their potential and limitations in VFR tasks. In addition, we studied the effectiveness of the commonly used image quality assessment (IQA) metrics and face IQA (FIQA) metrics by leveraging a subjective user study. With extensive experimental results and detailed analysis provided, we gained insights from the successes and failures of both current BFR and VSR methods. These results also pose challenges to current face restoration approaches, which we hope stimulate future advances in VFR research.
Authors: Sho Ueda, Xujun Ye
Abstract: Accurately assessing tomato plant nutrient status is crucial for maintaining high yields. Consequently, accurately identifying fertilizer-induced stress through the morphological traits of tomato plants has become a critical agricultural challenge. Research and development efforts have focused on developing noninvasive diagnostic tools for nutrition that leverage a combination of morphological traits and advanced sensor technologies. Given these advancements, detecting fertilizer stress by observing morphological traits near the growth points of tomatoes is still a significant challenge. To address this challenge, we developed a simple and cost-effective smartphone-based method for measuring trichome density. This method involves transferring trichomes from the surface of a leaf onto cellophane tape and capturing images using a smartphone. The images are processed using computer vision techniques to calculate the trichome density. To assess the efficacy of this method, we performed experiments on hydroponically grown tomato plants subjected to varying fertilizer concentrations. Our results indicate that our novel method for measuring trichome density accurately reflects fertilizer stress in tomato plants. The predictive performance of our model, as evaluated by the mean area under the precision recall curve, was 0.824, despite variations in the measurement data caused by differences in optical conditions. This study introduces an innovative approach for designing diagnostic devices for detecting fertilizer stress in plants by considering the surface structures of plants. Our proposed method represents a straightforward, efficient, and economical approach for evaluating the nutrient status of tomato plants and has the potential to overcome the limitations of conventional noncontact optical methods.
Authors: Luxi Chen, Zhengyi Wang, Chongxuan Li, Tingting Gao, Hang Su, Jun Zhu
Abstract: Optimization-based approaches, such as score distillation sampling (SDS), show promise in zero-shot 3D generation but suffer from low efficiency, primarily due to the high number of function evaluations (NFEs) required for each sample. In this paper, we introduce score-based iterative reconstruction (SIR), an efficient and general algorithm for 3D generation with a multi-view score-based diffusion model. Given the images produced by the diffusion model, SIR reduces NFEs by repeatedly optimizing 3D parameters, unlike the single optimization in SDS, mimicking the 3D reconstruction process. With other improvements including optimization in the pixel space, we present an efficient approach called MicroDreamer that generally applies to various 3D representations and 3D generation tasks. In particular, retaining a comparable performance, MicroDreamer is 5-20 times faster than SDS in generating neural radiance field and takes about 20 seconds to generate meshes from 3D Gaussian splitting on a single A100 GPU, halving the time of the fastest zero-shot baseline, DreamGaussian. Our code is available at https://github.com/ML-GSAI/MicroDreamer.
Authors: Yunbing Jia, Xiaoyu Kong, Fan Tang, Yixing Gao, Weiming Dong, Yi Yang
Abstract: In this paper, we reveal the two sides of data augmentation: enhancements in closed-set recognition correlate with a significant decrease in open-set recognition. Through empirical investigation, we find that multi-sample-based augmentations would contribute to reducing feature discrimination, thereby diminishing the open-set criteria. Although knowledge distillation could impair the feature via imitation, the mixed feature with ambiguous semantics hinders the distillation. To this end, we propose an asymmetric distillation framework by feeding teacher model extra raw data to enlarge the benefit of teacher. Moreover, a joint mutual information loss and a selective relabel strategy are utilized to alleviate the influence of hard mixed samples. Our method successfully mitigates the decline in open-set and outperforms SOTAs by 2%~3% AUROC on the Tiny-ImageNet dataset and experiments on large-scale dataset ImageNet-21K demonstrate the generalization of our method.
Authors: Norman Mu, Jingwei Ji, Zhenpei Yang, Nate Harada, Haotian Tang, Kan Chen, Charles R. Qi, Runzhou Ge, Kratarth Goel, Zoey Yang, Scott Ettinger, Rami Al-Rfou, Dragomir Anguelov, Yin Zhou
Abstract: Many existing motion prediction approaches rely on symbolic perception outputs to generate agent trajectories, such as bounding boxes, road graph information and traffic lights. This symbolic representation is a high-level abstraction of the real world, which may render the motion prediction model vulnerable to perception errors (e.g., failures in detecting open-vocabulary obstacles) while missing salient information from the scene context (e.g., poor road conditions). An alternative paradigm is end-to-end learning from raw sensors. However, this approach suffers from the lack of interpretability and requires significantly more training resources. In this work, we propose tokenizing the visual world into a compact set of scene elements and then leveraging pre-trained image foundation models and LiDAR neural networks to encode all the scene elements in an open-vocabulary manner. The image foundation model enables our scene tokens to encode the general knowledge of the open world while the LiDAR neural network encodes geometry information. Our proposed representation can efficiently encode the multi-frame multi-modality observations with a few hundred tokens and is compatible with most transformer-based architectures. To evaluate our method, we have augmented Waymo Open Motion Dataset with camera embeddings. Experiments over Waymo Open Motion Dataset show that our approach leads to significant performance improvements over the state-of-the-art.
Authors: Yuekun Dai, Dafeng Zhang, Xiaoming Li, Zongsheng Yue, Chongyi Li, Shangchen Zhou, Ruicheng Feng, Peiqing Yang, Zhezhu Jin, Guanqun Liu, Chen Change Loy, Lize Zhang, Shuai Liu, Chaoyu Feng, Luyang Wang, Shuan Chen, Guangqi Shao, Xiaotao Wang, Lei Lei, Qirui Yang, Qihua Cheng, Zhiqiang Xu, Yihao Liu, Huanjing Yue, Jingyu Yang, Florin-Alexandru Vasluianu, Zongwei Wu, George Ciubotariu, Radu Timofte, Zhao Zhang, Suiyi Zhao, Bo Wang, Zhichao Zuo, Yanyan Wei, Kuppa Sai Sri Teja, Jayakar Reddy A, Girish Rongali, Kaushik Mitra, Zhihao Ma, Yongxu Liu, Wanying Zhang, Wei Shang, Yuhong He, Long Peng, Zhongxin Yu, Shaofei Luo, Jian Wang, Yuqi Miao, Baiang Li, Gang Wei, Rakshank Verma, Ritik Maheshwari, Rahul Tekchandani, Praful Hambarde, Satya Narayan Tazi, Santosh Kumar Vipparthi, Subrahmanyam Murala, Haopeng Zhang, Yingli Hou, Mingde Yao, Levin M S, Aniruth Sundararajan, Hari Kumar A
Abstract: The increasing demand for computational photography and imaging on mobile platforms has led to the widespread development and integration of advanced image sensors with novel algorithms in camera systems. However, the scarcity of high-quality data for research and the rare opportunity for in-depth exchange of views from industry and academia constrain the development of mobile intelligent photography and imaging (MIPI). Building on the achievements of the previous MIPI Workshops held at ECCV 2022 and CVPR 2023, we introduce our third MIPI challenge including three tracks focusing on novel image sensors and imaging algorithms. In this paper, we summarize and review the Nighttime Flare Removal track on MIPI 2024. In total, 170 participants were successfully registered, and 14 teams submitted results in the final testing phase. The developed solutions in this challenge achieved state-of-the-art performance on Nighttime Flare Removal. More details of this challenge and the link to the dataset can be found at https://mipi-challenge.org/MIPI2024/.
Authors: Rayan Armani, Changlin Qian, Jiaxi Jiang, Christian Holz
Abstract: While camera-based capture systems remain the gold standard for recording human motion, learning-based tracking systems based on sparse wearable sensors are gaining popularity. Most commonly, they use inertial sensors, whose propensity for drift and jitter have so far limited tracking accuracy. In this paper, we propose Ultra Inertial Poser, a novel 3D full body pose estimation method that constrains drift and jitter in inertial tracking via inter-sensor distances. We estimate these distances across sparse sensor setups using a lightweight embedded tracker that augments inexpensive off-the-shelf 6D inertial measurement units with ultra-wideband radio-based ranging$-$dynamically and without the need for stationary reference anchors. Our method then fuses these inter-sensor distances with the 3D states estimated from each sensor Our graph-based machine learning model processes the 3D states and distances to estimate a person's 3D full body pose and translation. To train our model, we synthesize inertial measurements and distance estimates from the motion capture database AMASS. For evaluation, we contribute a novel motion dataset of 10 participants who performed 25 motion types, captured by 6 wearable IMU+UWB trackers and an optical motion capture system, totaling 200 minutes of synchronized sensor data (UIP-DB). Our extensive experiments show state-of-the-art performance for our method over PIP and TIP, reducing position error from $13.62$ to $10.65cm$ ($22\%$ better) and lowering jitter from $1.56$ to $0.055km/s^3$ (a reduction of $97\%$).
Authors: Trung Thanh Nguyen, Yasutomo Kawanishi, Takahiro Komamizu, Ichiro Ide
Abstract: Open-vocabulary Temporal Action Detection (Open-vocab TAD) is an advanced video analysis approach that expands Closed-vocabulary Temporal Action Detection (Closed-vocab TAD) capabilities. Closed-vocab TAD is typically confined to localizing and classifying actions based on a predefined set of categories. In contrast, Open-vocab TAD goes further and is not limited to these predefined categories. This is particularly useful in real-world scenarios where the variety of actions in videos can be vast and not always predictable. The prevalent methods in Open-vocab TAD typically employ a 2-stage approach, which involves generating action proposals and then identifying those actions. However, errors made during the first stage can adversely affect the subsequent action identification accuracy. Additionally, existing studies face challenges in handling actions of different durations owing to the use of fixed temporal processing methods. Therefore, we propose a 1-stage approach consisting of two primary modules: Multi-scale Video Analysis (MVA) and Video-Text Alignment (VTA). The MVA module captures actions at varying temporal resolutions, overcoming the challenge of detecting actions with diverse durations. The VTA module leverages the synergy between visual and textual modalities to precisely align video segments with corresponding action labels, a critical step for accurate action identification in Open-vocab scenarios. Evaluations on widely recognized datasets THUMOS14 and ActivityNet-1.3, showed that the proposed method achieved superior results compared to the other methods in both Open-vocab and Closed-vocab settings. This serves as a strong demonstration of the effectiveness of the proposed method in the TAD task.
Authors: Lei Wang, Desen Yuan
Abstract: Despite great success in modeling visual perception, deep neural network based image quality assessment (IQA) still remains unreliable in real-world applications due to its vulnerability to adversarial perturbations and the inexplicit black-box structure. In this paper, we propose to build a trustworthy IQA model via Causal Perception inspired Representation Learning (CPRL), and a score reflection attack method for IQA model. More specifically, we assume that each image is composed of Causal Perception Representation (CPR) and non-causal perception representation (N-CPR). CPR serves as the causation of the subjective quality label, which is invariant to the imperceptible adversarial perturbations. Inversely, N-CPR presents spurious associations with the subjective quality label, which may significantly change with the adversarial perturbations. To extract the CPR from each input image, we develop a soft ranking based channel-wise activation function to mediate the causally sufficient (beneficial for high prediction accuracy) and necessary (beneficial for high robustness) deep features, and based on intervention employ minimax game to optimize. Experiments on four benchmark databases show that the proposed CPRL method outperforms many state-of-the-art adversarial defense methods and provides explicit model interpretation.
Authors: Adewale Emmanuel Adedokun, Muhammed Bashir Abdulrazak, Muyideen Momoh Omuya, Habeeb BelloSalau, Bashir Olaniyi Sadiq
Abstract: Real-time video life streaming of events over a network continued to gain more popularity among the populace. However, there is need to ensure the judicious utilization of allocated bandwidth without compromising the Quality of Service (QoS) of the system. In this regard, this paper presents an approach based on spatio-temporal frame indexing that detects and eliminate redundancy within and across captured frame, prior transmission from the server to clients. The standard and local low motion videos were the two scenarios considered in evaluating the performance of the proposed algorithm. Results obtained showed that the proposed approach achieved an improvement of 5.13%, 15.8% and 5%, 15.6% improvement in terms of the buffer size and compression ratio. Though with a tradeoff of the frame-built time, where both the standard and local frame indexing outperforms the proposed scheme with 10.8% and 8.71% respectively.
Authors: Francesca Razzano, Pietro Di Stasio, Francesco Mauro, Gabriele Meoni, Marco Esposito, Gilda Schirinzi, Silvia L. Ullo
Abstract: Differently from conventional procedures, the proposed solution advocates for a groundbreaking paradigm in water quality monitoring through the integration of satellite Remote Sensing (RS) data, Artificial Intelligence (AI) techniques, and onboard processing. The objective is to offer nearly real-time detection of contaminants in coastal waters addressing a significant gap in the existing literature. Moreover, the expected outcomes include substantial advancements in environmental monitoring, public health protection, and resource conservation. The specific focus of our study is on the estimation of Turbidity and pH parameters, for their implications on human and aquatic health. Nevertheless, the designed framework can be extended to include other parameters of interest in the water environment and beyond. Originating from our participation in the European Space Agency (ESA) OrbitalAI Challenge, this article describes the distinctive opportunities and issues for the contaminants monitoring on the Phisat-2 mission. The specific characteristics of this mission, with the tools made available, will be presented, with the methodology proposed by the authors for the onboard monitoring of water contaminants in near real-time. Preliminary promising results are discussed and in progress and future work introduced.
Authors: Lei Wang, Desen Yuan
Abstract: In this paper, we propose a highly efficient method to estimate an image's mean opinion score (MOS) from a single opinion score (SOS). Assuming that each SOS is the observed sample of a normal distribution and the MOS is its unknown expectation, the MOS inference is formulated as a maximum likelihood estimation problem, where the perceptual correlation of pairwise images is considered in modeling the likelihood of SOS. More specifically, by means of the quality-aware representations learned from the self-supervised backbone, we introduce a learnable relative quality measure to predict the MOS difference between two images. Then, the current image's maximum likelihood estimation towards MOS is represented by the sum of another reference image's estimated MOS and their relative quality. Ideally, no matter which image is selected as the reference, the MOS of the current image should remain unchanged, which is termed perceptual cons tancy constrained calibration (PC3). Finally, we alternatively optimize the relative quality measure's parameter and the current image's estimated MOS via backpropagation and Newton's method respectively. Experiments show that the proposed method is efficient in calibrating the biased SOS and significantly improves IQA model learning when only SOSs are available.
Authors: Denys Godwin, Hanxi Li, Michael Cecil, Hamed Alemohammad
Abstract: Filling cloudy pixels in multispectral satellite imagery is essential for accurate data analysis and downstream applications, especially for tasks which require time series data. To address this issue, we compare the performance of a foundational Vision Transformer (ViT) model with a baseline Conditional Generative Adversarial Network (CGAN) model for missing value imputation in time series of multispectral satellite imagery. We randomly mask time series of satellite images using real-world cloud masks and train each model to reconstruct the missing pixels. The ViT model is fine-tuned from a pretrained model, while the CGAN is trained from scratch. Using quantitative evaluation metrics such as structural similarity index and mean absolute error as well as qualitative visual analysis, we assess imputation accuracy and contextual preservation.
Authors: Yue Li, Baiqiao Yin, Jinfu Liu, Jiajun Wen, Jiaying Lin, Mengyuan Liu
Abstract: In recent years, Event Sound Source Localization has been widely applied in various fields. Recent works typically relying on the contrastive learning framework show impressive performance. However, all work is based on large relatively simple datasets. It's also crucial to understand and analyze human behaviors (actions and interactions of people), voices, and sounds in chaotic events in many applications, e.g., crowd management, and emergency response services. In this paper, we apply the existing model to a more complex dataset, explore the influence of parameters on the model, and propose a semi-supervised improvement method SemiPL. With the increase in data quantity and the influence of label quality, self-supervised learning will be an unstoppable trend. The experiment shows that the parameter adjustment will positively affect the existing model. In particular, SSPL achieved an improvement of 12.2% cIoU and 0.56% AUC in Chaotic World compared to the results provided. The code is available at: https://github.com/ly245422/SSPL
Authors: Jiayi Han, Zidi Cao, Weibo Zheng, Xiangguo Zhou, Xiangjian He, Yuanfang Zhang, Daisen Wei
Abstract: In recent years, zero-shot learning has attracted the focus of many researchers, due to its flexibility and generality. Many approaches have been proposed to achieve the zero-shot classification of the point clouds for 3D object understanding, following the schema of CLIP. However, in the real world, the point clouds could be extremely sparse, dramatically limiting the effectiveness of the 3D point cloud encoders, and resulting in the misalignment of point cloud features and text embeddings. To the point cloud encoders to fit the extremely sparse point clouds without re-running the pre-training procedure which could be time-consuming and expensive, in this work, we propose an unsupervised model adaptation approach to enhance the point cloud encoder for the extremely sparse point clouds. We propose a novel fused-cross attention layer that expands the pre-trained self-attention layer with additional learnable tokens and attention blocks, which effectively modifies the point cloud features while maintaining the alignment between point cloud features and text embeddings. We also propose a complementary learning-based self-distillation schema that encourages the modified features to be pulled apart from the irrelevant text embeddings without overfitting the feature space to the observed text embeddings. Extensive experiments demonstrate that the proposed approach effectively increases the zero-shot capability on extremely sparse point clouds, and overwhelms other state-of-the-art model adaptation approaches.
Authors: Min Zhang, Haoxuan Li, Fei Wu, Kun Kuang
Abstract: Out-of-distribution (OOD) problems in few-shot classification (FSC) occur when novel classes sampled from testing distributions differ from base classes drawn from training distributions, which considerably degrades the performance of deep learning models deployed in real-world applications. Recent studies suggest that the OOD problems in FSC mainly including: (a) cross-domain few-shot classification (CD-FSC) and (b) spurious-correlation few-shot classification (SC-FSC). Specifically, CD-FSC occurs when a classifier learns transferring knowledge from base classes drawn from seen training distributions but recognizes novel classes sampled from unseen testing distributions. In contrast, SC-FSC arises when a classifier relies on non-causal features (or contexts) that happen to be correlated with the labels (or concepts) in base classes but such relationships no longer hold during the model deployment. Despite CD-FSC has been extensively studied, SC-FSC remains understudied due to lack of the corresponding evaluation benchmarks. To this end, we present Meta Concept Context (MetaCoCo), a benchmark with spurious-correlation shifts collected from real-world scenarios. Moreover, to quantify the extent of spurious-correlation shifts of the presented MetaCoCo, we further propose a metric by using CLIP as a pre-trained vision-language model. Extensive experiments on the proposed benchmark are performed to evaluate the state-of-the-art methods in FSC, cross-domain shifts, and self-supervised learning. The experimental results show that the performance of the existing methods degrades significantly in the presence of spurious-correlation shifts. We open-source all codes of our benchmark and hope that the proposed MetaCoCo can facilitate future research on spurious-correlation shifts problems in FSC. The code is available at: https://github.com/remiMZ/MetaCoCo-ICLR24.
Authors: Yuliang Liu, Mingxin Huang, Hao Yan, Linger Deng, Weijia Wu, Hao Lu, Chunhua Shen, Lianwen Jin, Xiang Bai
Abstract: Text spotting, a task involving the extraction of textual information from image or video sequences, faces challenges in cross-domain adaption, such as image-to-image and image-to-video generalization. In this paper, we introduce a new method, termed VimTS, which enhances the generalization ability of the model by achieving better synergy among different tasks. Typically, we propose a Prompt Queries Generation Module and a Tasks-aware Adapter to effectively convert the original single-task model into a multi-task model suitable for both image and video scenarios with minimal additional parameters. The Prompt Queries Generation Module facilitates explicit interaction between different tasks, while the Tasks-aware Adapter helps the model dynamically learn suitable features for each task. Additionally, to further enable the model to learn temporal information at a lower cost, we propose a synthetic video text dataset (VTD-368k) by leveraging the Content Deformation Fields (CoDeF) algorithm. Notably, our method outperforms the state-of-the-art method by an average of 2.6% in six cross-domain benchmarks such as TT-to-IC15, CTW1500-to-TT, and TT-to-CTW1500. For video-level cross-domain adaption, our method even surpasses the previous end-to-end video spotting method in ICDAR2015 video and DSText v2 by an average of 5.5% on the MOTA metric, using only image-level data. We further demonstrate that existing Large Multimodal Models exhibit limitations in generating cross-domain scene text spotting, in contrast to our VimTS model which requires significantly fewer parameters and data. The code and datasets will be made available at the https://VimTextSpotter.github.io.
Authors: Rishav Pramanik, Jos\'e-Fabian Villa-V\'asquez, Marco Pedersoli
Abstract: Unsupervised object discovery is becoming an essential line of research for tackling recognition problems that require decomposing an image into entities, such as semantic segmentation and object detection. Recently, object-centric methods that leverage self-supervision have gained popularity, due to their simplicity and adaptability to different settings and conditions. However, those methods do not exploit effective techniques already employed in modern self-supervised approaches. In this work, we consider an object-centric approach in which DINO ViT features are reconstructed via a set of queried representations called slots. Based on that, we propose a masking scheme on input features that selectively disregards the background regions, inducing our model to focus more on salient objects during the reconstruction phase. Moreover, we extend the slot attention to a multi-query approach, allowing the model to learn multiple sets of slots, producing more stable masks. During training, these multiple sets of slots are learned independently while, at test time, these sets are merged through Hungarian matching to obtain the final slots. Our experimental results and ablations on the PASCAL-VOC 2012 dataset show the importance of each component and highlight how their combination consistently improves object localization. Our source code is available at: https://github.com/rishavpramanik/maskedmultiqueryslot
URLs: https://github.com/rishavpramanik/maskedmultiqueryslot
Authors: Felix Gr\"un, Marcus Nolte, Markus Maurer
Abstract: The foundational role of datasets in defining the capabilities of deep learning models has led to their rapid proliferation. At the same time, published research focusing on the process of dataset development for environment perception in automated driving has been scarce, thereby reducing the applicability of openly available datasets and impeding the development of effective environment perception systems. Sensor-based, mapless automated driving is one of the contexts where this limitation is evident. While leveraging real-time sensor data, instead of pre-defined HD maps promises enhanced adaptability and safety by effectively navigating unexpected environmental changes, it also increases the demands on the scope and complexity of the information provided by the perception system. To address these challenges, we propose a scenario- and capability-based approach to dataset development. Grounded in the principles of ISO 21448 (safety of the intended functionality, SOTIF), extended by ISO/TR 4804, our approach facilitates the structured derivation of dataset requirements. This not only aids in the development of meaningful new datasets but also enables the effective comparison of existing ones. Applying this methodology to a broad range of existing lane detection datasets, we identify significant limitations in current datasets, particularly in terms of real-world applicability, a lack of labeling of critical features, and an absence of comprehensive information for complex driving maneuvers.
Authors: Lei Wang, Desen Yuan
Abstract: Image quality assessment often relies on raw opinion scores provided by subjects in subjective experiments, which can be noisy and unreliable. To address this issue, postprocessing procedures such as ITU-R BT.500, ITU-T P.910, and ITU-T P.913 have been standardized to clean up the original opinion scores. These methods use annotator-based statistical priors, but they do not take into account extensive information about the image itself, which limits their performance in less annotated scenarios. Generally speaking, image quality datasets usually contain similar scenes or distortions, and it is inevitable for subjects to compare images to score a reasonable score when scoring. Therefore, In this paper, we proposed Subjective Image Quality Score Preprocessing Method perceptual similarity Subjective Preprocessing (PSP), which exploit the perceptual similarity between images to alleviate subjective bias in less annotated scenarios. Specifically, we model subjective scoring as a conditional probability model based on perceptual similarity with previously scored images, called subconscious reference scoring. The reference images are stored by a neighbor dictionary, which is obtained by a normalized vector dot-product based nearest neighbor search of the images' perceptual depth features. Then the preprocessed score is updated by the exponential moving average (EMA) of the subconscious reference scoring, called similarity regularized EMA. Our experiments on multiple datasets (LIVE, TID2013, CID2013) show that this method can effectively remove the bias of the subjective scores. Additionally, Experiments prove that the Preprocesed dataset can improve the performance of downstream IQA tasks very well.
Authors: Chun Feng, Joy Hsu, Weiyu Liu, Jiajun Wu
Abstract: 3D visual grounding is a challenging task that often requires direct and dense supervision, notably the semantic label for each object in the scene. In this paper, we instead study the naturally supervised setting that learns from only 3D scene and QA pairs, where prior works underperform. We propose the Language-Regularized Concept Learner (LARC), which uses constraints from language as regularization to significantly improve the accuracy of neuro-symbolic concept learners in the naturally supervised setting. Our approach is based on two core insights: the first is that language constraints (e.g., a word's relation to another) can serve as effective regularization for structured representations in neuro-symbolic models; the second is that we can query large language models to distill such constraints from language properties. We show that LARC improves performance of prior works in naturally supervised 3D visual grounding, and demonstrates a wide range of 3D visual reasoning capabilities-from zero-shot composition, to data efficiency and transferability. Our method represents a promising step towards regularizing structured visual reasoning frameworks with language-based priors, for learning in settings without dense supervision.
Authors: Kai Zhang, Sai Bi, Hao Tan, Yuanbo Xiangli, Nanxuan Zhao, Kalyan Sunkavalli, Zexiang Xu
Abstract: We propose GS-LRM, a scalable large reconstruction model that can predict high-quality 3D Gaussian primitives from 2-4 posed sparse images in 0.23 seconds on single A100 GPU. Our model features a very simple transformer-based architecture; we patchify input posed images, pass the concatenated multi-view image tokens through a sequence of transformer blocks, and decode final per-pixel Gaussian parameters directly from these tokens for differentiable rendering. In contrast to previous LRMs that can only reconstruct objects, by predicting per-pixel Gaussians, GS-LRM naturally handles scenes with large variations in scale and complexity. We show that our model can work on both object and scene captures by training it on Objaverse and RealEstate10K respectively. In both scenarios, the models outperform state-of-the-art baselines by a wide margin. We also demonstrate applications of our model in downstream 3D generation tasks. Our project webpage is available at: https://sai-bi.github.io/project/gs-lrm/ .
Authors: Zhexi Peng, Tianjia Shao, Yong Liu, Jingke Zhou, Yin Yang, Jingdong Wang, Kun Zhou
Abstract: We propose RTG-SLAM, a real-time 3D reconstruction system with an RGBD camera for large-scale environments using Gaussian splatting. RTG-SLAM features a compact Gaussian representation and a highly efficient on-the-fly Gaussian optimization scheme. We force each Gaussian to be either opaque or nearly transparent, with the opaque ones fitting the surface and dominant colors, and transparent ones fitting residual colors. By rendering depth in a different way from color rendering, we let a single opaque Gaussian well fit a local surface region without the need of multiple overlapping Gaussians, hence largely reducing the memory and computation cost. For on-the-fly Gaussian optimization, we explicitly add Gaussians for three types of pixels per frame: newly observed, with large color errors and with large depth errors. We also categorize all Gaussians into stable and unstable ones, where the stable Gaussians are expected to well fit previously observed RGBD images and otherwise unstable. We only optimize the unstable Gaussians and only render the pixels occupied by unstable Gaussians. In this way, both the number of Gaussians to be optimized and pixels to be rendered are largely reduced, and the optimization can be done in real time. We show real-time reconstructions of a variety of real large scenes. Compared with the state-of-the-art NeRF-based RGBD SLAM, our system achieves comparable high-quality reconstruction but with around twice the speed and half the memory cost, and shows superior performance in the realism of novel view synthesis and camera tracking accuracy.
Authors: Jingbo Wang, Zhengyi Luo, Ye Yuan, Yixuan Li, Bo Dai
Abstract: We address the challenge of content diversity and controllability in pedestrian simulation for driving scenarios. Recent pedestrian animation frameworks have a significant limitation wherein they primarily focus on either following trajectory [46] or the content of the reference video [57], consequently overlooking the potential diversity of human motion within such scenarios. This limitation restricts the ability to generate pedestrian behaviors that exhibit a wider range of variations and realistic motions and therefore restricts its usage to provide rich motion content for other components in the driving simulation system, e.g., suddenly changed motion to which the autonomous vehicle should respond. In our approach, we strive to surpass the limitation by showcasing diverse human motions obtained from various sources, such as generated human motions, in addition to following the given trajectory. The fundamental contribution of our framework lies in combining the motion tracking task with trajectory following, which enables the tracking of specific motion parts (e.g., upper body) while simultaneously following the given trajectory by a single policy. This way, we significantly enhance both the diversity of simulated human motion within the given scenario and the controllability of the content, including language-based control. Our framework facilitates the generation of a wide range of human motions, contributing to greater realism and adaptability in pedestrian simulations for driving scenarios. More information is on our project page https://wangjingbo1219.github.io/papers/CVPR2024_PACER_PLUS/PACERPLUSPage.html .
URLs: https://wangjingbo1219.github.io/papers/CVPR2024_PACER_PLUS/PACERPLUSPage.html
Authors: Zhipeng Yuan, Nasamu Musa, Katarzyna Dybal, Matthew Back, Daniel Leybourne, Po Yang
Abstract: Every year, plant parasitic nematodes, one of the major groups of plant pathogens, cause a significant loss of crops worldwide. To mitigate crop yield losses caused by nematodes, an efficient nematode monitoring method is essential for plant and crop disease management. In other respects, efficient nematode detection contributes to medical research and drug discovery, as nematodes are model organisms. With the rapid development of computer technology, computer vision techniques provide a feasible solution for quantifying nematodes or nematode infections. In this paper, we survey and categorise the studies and available datasets on nematode detection through deep-learning models. To stimulate progress in related research, this survey presents the potential state-of-the-art object detection models, training techniques, optimisation techniques, and evaluation metrics for deep learning beginners. Moreover, seven state-of-the-art object detection models are validated on three public datasets and the AgriNema dataset for plant parasitic nematodes to construct a baseline for nematode detection.
Authors: Yunhao Ge, Xiaohui Zeng, Jacob Samuel Huffman, Tsung-Yi Lin, Ming-Yu Liu, Yin Cui
Abstract: Existing automatic captioning methods for visual content face challenges such as lack of detail, content hallucination, and poor instruction following. In this work, we propose VisualFactChecker (VFC), a flexible training-free pipeline that generates high-fidelity and detailed captions for both 2D images and 3D objects. VFC consists of three steps: 1) proposal, where image-to-text captioning models propose multiple initial captions; 2) verification, where a large language model (LLM) utilizes tools such as object detection and VQA models to fact-check proposed captions; 3) captioning, where an LLM generates the final caption by summarizing caption proposals and the fact check verification results. In this step, VFC can flexibly generate captions in various styles following complex instructions. We conduct comprehensive captioning evaluations using four metrics: 1) CLIP-Score for image-text similarity; 2) CLIP-Image-Score for measuring the image-image similarity between the original and the reconstructed image generated by a text-to-image model using the caption. 3) human study on Amazon Mechanical Turk; 4) GPT-4V for fine-grained evaluation. Evaluation results show that VFC outperforms state-of-the-art open-sourced captioning methods for 2D images on the COCO dataset and 3D assets on the Objaverse dataset. Our study demonstrates that by combining open-source models into a pipeline, we can attain captioning capability comparable to proprietary models such as GPT-4V, despite being over 10x smaller in model size.
Authors: Yasumasa Onoe, Sunayana Rane, Zachary Berger, Yonatan Bitton, Jaemin Cho, Roopal Garg, Alexander Ku, Zarana Parekh, Jordi Pont-Tuset, Garrett Tanzer, Su Wang, Jason Baldridge
Abstract: Vision-language datasets are vital for both text-to-image (T2I) and image-to-text (I2T) research. However, current datasets lack descriptions with fine-grained detail that would allow for richer associations to be learned by models. To fill the gap, we introduce Descriptions of Connected and Contrasting Images (DOCCI), a dataset with long, human-annotated English descriptions for 15k images that were taken, curated and donated by a single researcher intent on capturing key challenges such as spatial relations, counting, text rendering, world knowledge, and more. We instruct human annotators to create comprehensive descriptions for each image; these average 136 words in length and are crafted to clearly distinguish each image from those that are related or similar. Each description is highly compositional and typically encompasses multiple challenges. Through both quantitative and qualitative analyses, we demonstrate that DOCCI serves as an effective training resource for image-to-text generation -- a PaLI 5B model finetuned on DOCCI shows equal or superior results compared to highly-performant larger models like LLaVA-1.5 7B and InstructBLIP 7B. Furthermore, we show that DOCCI is a useful testbed for text-to-image generation, highlighting the limitations of current text-to-image models in capturing long descriptions and fine details.
Authors: Paul Engstler, Andrea Vedaldi, Iro Laina, Christian Rupprecht
Abstract: 3D scene generation has quickly become a challenging new research direction, fueled by consistent improvements of 2D generative diffusion models. Most prior work in this area generates scenes by iteratively stitching newly generated frames with existing geometry. These works often depend on pre-trained monocular depth estimators to lift the generated images into 3D, fusing them with the existing scene representation. These approaches are then often evaluated via a text metric, measuring the similarity between the generated images and a given text prompt. In this work, we make two fundamental contributions to the field of 3D scene generation. First, we note that lifting images to 3D with a monocular depth estimation model is suboptimal as it ignores the geometry of the existing scene. We thus introduce a novel depth completion model, trained via teacher distillation and self-training to learn the 3D fusion process, resulting in improved geometric coherence of the scene. Second, we introduce a new benchmarking scheme for scene generation methods that is based on ground truth geometry, and thus measures the quality of the structure of the scene.
Authors: Wenxun Dai, Ling-Hao Chen, Jingbo Wang, Jinpeng Liu, Bo Dai, Yansong Tang
Abstract: This work introduces MotionLCM, extending controllable motion generation to a real-time level. Existing methods for spatial control in text-conditioned motion generation suffer from significant runtime inefficiency. To address this issue, we first propose the motion latent consistency model (MotionLCM) for motion generation, building upon the latent diffusion model (MLD). By employing one-step (or few-step) inference, we further improve the runtime efficiency of the motion latent diffusion model for motion generation. To ensure effective controllability, we incorporate a motion ControlNet within the latent space of MotionLCM and enable explicit control signals (e.g., pelvis trajectory) in the vanilla motion space to control the generation process directly, similar to controlling other latent-free diffusion models for motion generation. By employing these techniques, our approach can generate human motions with text and control signals in real-time. Experimental results demonstrate the remarkable generation and controlling capabilities of MotionLCM while maintaining real-time runtime efficiency.
Authors: Ang Cao, Justin Johnson, Andrea Vedaldi, David Novotny
Abstract: Contemporary 3D research, particularly in reconstruction and generation, heavily relies on 2D images for inputs or supervision. However, current designs for these 2D-3D mapping are memory-intensive, posing a significant bottleneck for existing methods and hindering new applications. In response, we propose a pair of highly scalable components for 3D neural fields: Lightplane Render and Splatter, which significantly reduce memory usage in 2D-3D mapping. These innovations enable the processing of vastly more and higher resolution images with small memory and computational costs. We demonstrate their utility in various applications, from benefiting single-scene optimization with image-level losses to realizing a versatile pipeline for dramatically scaling 3D reconstruction and generation. Code: \url{https://github.com/facebookresearch/lightplane}.
Authors: Jun Yu, Yutong Dai, Xiaokang Liu, Jin Huang, Yishan Shen, Ke Zhang, Rong Zhou, Eashan Adhikarla, Wenxuan Ye, Yixin Liu, Zhaoming Kong, Kai Zhang, Yilong Yin, Vinod Namboodiri, Brian D. Davison, Jason H. Moore, Yong Chen
Abstract: MTL is a learning paradigm that effectively leverages both task-specific and shared information to address multiple related tasks simultaneously. In contrast to STL, MTL offers a suite of benefits that enhance both the training process and the inference efficiency. MTL's key advantages encompass streamlined model architecture, performance enhancement, and cross-domain generalizability. Over the past twenty years, MTL has become widely recognized as a flexible and effective approach in various fields, including CV, NLP, recommendation systems, disease prognosis and diagnosis, and robotics. This survey provides a comprehensive overview of the evolution of MTL, encompassing the technical aspects of cutting-edge methods from traditional approaches to deep learning and the latest trend of pretrained foundation models. Our survey methodically categorizes MTL techniques into five key areas: regularization, relationship learning, feature propagation, optimization, and pre-training. This categorization not only chronologically outlines the development of MTL but also dives into various specialized strategies within each category. Furthermore, the survey reveals how the MTL evolves from handling a fixed set of tasks to embracing a more flexible approach free from task or modality constraints. It explores the concepts of task-promptable and -agnostic training, along with the capacity for ZSL, which unleashes the untapped potential of this historically coveted learning paradigm. Overall, we hope this survey provides the research community with a comprehensive overview of the advancements in MTL from its inception in 1997 to the present in 2023. We address present challenges and look ahead to future possibilities, shedding light on the opportunities and potential avenues for MTL research in a broad manner. This project is publicly available at https://github.com/junfish/Awesome-Multitask-Learning.
URLs: https://github.com/junfish/Awesome-Multitask-Learning.
Authors: Paul Pu Liang
Abstract: Building multisensory AI systems that learn from multiple sensory inputs such as text, speech, video, real-world sensors, wearable devices, and medical data holds great promise for impact in many scientific areas with practical benefits, such as in supporting human health and well-being, enabling multimedia content processing, and enhancing real-world autonomous agents. By synthesizing a range of theoretical frameworks and application domains, this thesis aims to advance the machine learning foundations of multisensory AI. In the first part, we present a theoretical framework formalizing how modalities interact with each other to give rise to new information for a task. These interactions are the basic building blocks in all multimodal problems, and their quantification enables users to understand their multimodal datasets, design principled approaches to learn these interactions, and analyze whether their model has succeeded in learning. In the second part, we study the design of practical multimodal foundation models that generalize over many modalities and tasks, which presents a step toward grounding large language models to real-world sensory modalities. We introduce MultiBench, a unified large-scale benchmark across a wide range of modalities, tasks, and research areas, followed by the cross-modal attention and multimodal transformer architectures that now underpin many of today's multimodal foundation models. Scaling these architectures on MultiBench enables the creation of general-purpose multisensory AI systems, and we discuss our collaborative efforts in applying these models for real-world impact in affective computing, mental health, cancer prognosis, and robotics. Finally, we conclude this thesis by discussing how future work can leverage these ideas toward more general, interactive, and safe multisensory AI.
Authors: Gabriel Sarch, Sahil Somani, Raghav Kapoor, Michael J. Tarr, Katerina Fragkiadaki
Abstract: Recent research on instructable agents has used memory-augmented Large Language Models (LLMs) as task planners, a technique that retrieves language-program examples relevant to the input instruction and uses them as in-context examples in the LLM prompt to improve the performance of the LLM in inferring the correct action and task plans. In this technical report, we extend the capabilities of HELPER, by expanding its memory with a wider array of examples and prompts, and by integrating additional APIs for asking questions. This simple expansion of HELPER into a shared memory enables the agent to work across the domains of executing plans from dialogue, natural language instruction following, active question asking, and commonsense room reorganization. We evaluate the agent on four diverse interactive visual-language embodied agent benchmarks: ALFRED, TEACh, DialFRED, and the Tidy Task. HELPER-X achieves few-shot, state-of-the-art performance across these benchmarks using a single agent, without requiring in-domain training, and remains competitive with agents that have undergone in-domain training.
Authors: K. Aditya Mohan, Massimiliano Ferrucci, Chuck Divin, Garrett A. Stevenson, Hyojin Kim
Abstract: 4D time-space reconstruction of dynamic events or deforming objects using X-ray computed tomography (CT) is an extremely ill-posed inverse problem. Existing approaches assume that the object remains static for the duration of several tens or hundreds of X-ray projection measurement images (reconstruction of consecutive limited-angle CT scans). However, this is an unrealistic assumption for many in-situ experiments that causes spurious artifacts and inaccurate morphological reconstructions of the object. To solve this problem, we propose to perform a 4D time-space reconstruction using a distributed implicit neural representation (DINR) network that is trained using a novel distributed stochastic training algorithm. Our DINR network learns to reconstruct the object at its output by iterative optimization of its network parameters such that the measured projection images best match the output of the CT forward measurement model. We use a continuous time and space forward measurement model that is a function of the DINR outputs at a sparsely sampled set of continuous valued object coordinates. Unlike existing state-of-the-art neural representation architectures that forward and back propagate through dense voxel grids that sample the object's entire time-space coordinates, we only propagate through the DINR at a small subset of object coordinates in each iteration resulting in an order-of-magnitude reduction in memory and compute for training. DINR leverages distributed computation across several compute nodes and GPUs to produce high-fidelity 4D time-space reconstructions even for extremely large CT data sizes. We use both simulated parallel-beam and experimental cone-beam X-ray CT datasets to demonstrate the superior performance of our approach.
Authors: Batuhan K. Karaman, Katerina Dodelzon, Gozde B. Akar, Mert R. Sabuncu
Abstract: Breast cancer is one of the leading causes of mortality among women worldwide. Early detection and risk assessment play a crucial role in improving survival rates. Therefore, annual or biennial mammograms are often recommended for screening in high-risk groups. Mammograms are typically interpreted by expert radiologists based on the Breast Imaging Reporting and Data System (BI-RADS), which provides a uniform way to describe findings and categorizes them to indicate the level of concern for breast cancer. Recently, machine learning (ML) and computational approaches have been developed to automate and improve the interpretation of mammograms. However, both BI-RADS and the ML-based methods focus on the analysis of data from the present and sometimes the most recent prior visit. While it is clear that temporal changes in image features of the longitudinal scans should carry value for quantifying breast cancer risk, no prior work has conducted a systematic study of this. In this paper, we extend a state-of-the-art ML model to ingest an arbitrary number of longitudinal mammograms and predict future breast cancer risk. On a large-scale dataset, we demonstrate that our model, LoMaR, achieves state-of-the-art performance when presented with only the present mammogram. Furthermore, we use LoMaR to characterize the predictive value of prior visits. Our results show that longer histories (e.g., up to four prior annual mammograms) can significantly boost the accuracy of predicting future breast cancer risk, particularly beyond the short-term. Our code and model weights are available at https://github.com/batuhankmkaraman/LoMaR.
Authors: Yipeng Zhang, Laurent Charlin, Richard Zemel, Mengye Ren
Abstract: We formulate a unifying framework for unsupervised continual learning (UCL), which disentangles learning objectives that are specific to the present and the past data, encompassing stability, plasticity, and cross-task consolidation. The framework reveals that many existing UCL approaches overlook cross-task consolidation and try to balance plasticity and stability in a shared embedding space. This results in worse performance due to a lack of within-task data diversity and reduced effectiveness in learning the current task. Our method, Osiris, which explicitly optimizes all three objectives on separate embedding spaces, achieves state-of-the-art performance on all benchmarks, including two novel benchmarks proposed in this paper featuring semantically structured task sequences. Compared to standard benchmarks, these two structured benchmarks more closely resemble visual signals received by humans and animals when navigating real-world environments. Finally, we show some preliminary evidence that continual models can benefit from such realistic learning scenarios.
Authors: Yao Gao, Qi Jiang, Shaohua Gao, Lei Sun, Kailun Yang, Kaiwei Wang
Abstract: The popularity of mobile vision creates a demand for advanced compact computational imaging systems, which call for the development of both a lightweight optical system and an effective image reconstruction model. Recently, joint design pipelines come to the research forefront, where the two significant components are simultaneously optimized via data-driven learning to realize the optimal system design. However, the effectiveness of these designs largely depends on the initial setup of the optical system, complicated by a non-convex solution space that impedes reaching a globally optimal solution. In this work, we present Global Search Optics (GSO) to automatically design compact computational imaging systems through two parts: (i) Fused Optimization Method for Automatic Optical Design (OptiFusion), which searches for diverse initial optical systems under certain design specifications; and (ii) Efficient Physic-aware Joint Optimization (EPJO), which conducts parallel joint optimization of initial optical systems and image reconstruction networks with the consideration of physical constraints, culminating in the selection of the optimal solution. Extensive experimental results on the design of three-piece (3P) sphere computational imaging systems illustrate that the GSO serves as a transformative end-to-end lens design paradigm for superior global optimal structure searching ability, which provides compact computational imaging systems with higher imaging quality compared to traditional methods. The source code will be made publicly available at https://github.com/wumengshenyou/GSO.
Authors: Wei Huang, Bingyang Zhang, Kaituo Zhang, Hua Gao, Rongchun Wan
Abstract: The task of anomaly detection is to separate anomalous data from normal data in the dataset. Models such as deep convolutional autoencoder (CAE) network and deep supporting vector data description (SVDD) model have been universally employed and have demonstrated significant success in detecting anomalies. However, the over-reconstruction ability of CAE network for anomalous data can easily lead to high false negative rate in detecting anomalous data. On the other hand, the deep SVDD model has the drawback of feature collapse, which leads to a decrease of detection accuracy for anomalies. To address these problems, we propose the Improved AutoEncoder with LSTM module and Kullback-Leibler divergence (IAE-LSTM-KL) model in this paper. An LSTM network is added after the encoder to memorize feature representations of normal data. In the meanwhile, the phenomenon of feature collapse can also be mitigated by penalizing the featured input to SVDD module via KL divergence. The efficacy of the IAE-LSTM-KL model is validated through experiments on both synthetic and real-world datasets. Experimental results show that IAE-LSTM-KL model yields higher detection accuracy for anomalies. In addition, it is also found that the IAE-LSTM-KL model demonstrates enhanced robustness to contaminated outliers in the dataset.
Authors: Shannon L. Walston, Hiroshi Seki, Hirotaka Takita, Yasuhito Mitsuyama, Shingo Sato, Akifumi Hagiwara, Rintaro Ito, Shouhei Hanaoka, Yukio Miki, Daiju Ueda
Abstract: Medicine and artificial intelligence (AI) engineering represent two distinct fields each with decades of published history. With such history comes a set of terminology that has a specific way in which it is applied. However, when two distinct fields with overlapping terminology start to collaborate, miscommunication and misunderstandings can occur. This narrative review aims to give historical context for these terms, accentuate the importance of clarity when these terms are used in medical AI contexts, and offer solutions to mitigate misunderstandings by readers from either field. Through an examination of historical documents, including articles, writing guidelines, and textbooks, this review traces the divergent evolution of terms for data sets and their impact. Initially, the discordant interpretations of the word 'validation' in medical and AI contexts are explored. Then the data sets used for AI evaluation are classified, namely random splitting, cross-validation, temporal, geographic, internal, and external sets. The accurate and standardized description of these data sets is crucial for demonstrating the robustness and generalizability of AI applications in medicine. This review clarifies existing literature to provide a comprehensive understanding of these classifications and their implications in AI evaluation. This review then identifies often misunderstood terms and proposes pragmatic solutions to mitigate terminological confusion. Among these solutions are the use of standardized terminology such as 'training set,' 'validation (or tuning) set,' and 'test set,' and explicit definition of data set splitting terminologies in each medical AI research publication. This review aspires to enhance the precision of communication in medical AI, thereby fostering more effective and transparent research methodologies in this interdisciplinary field.
Authors: Shengjie Ma, Yanlin Weng, Tianjia Shao, Kun Zhou
Abstract: We introduce 3D Gaussian blendshapes for modeling photorealistic head avatars. Taking a monocular video as input, we learn a base head model of neutral expression, along with a group of expression blendshapes, each of which corresponds to a basis expression in classical parametric face models. Both the neutral model and expression blendshapes are represented as 3D Gaussians, which contain a few properties to depict the avatar appearance. The avatar model of an arbitrary expression can be effectively generated by combining the neutral model and expression blendshapes through linear blending of Gaussians with the expression coefficients. High-fidelity head avatar animations can be synthesized in real time using Gaussian splatting. Compared to state-of-the-art methods, our Gaussian blendshape representation better captures high-frequency details exhibited in input video, and achieves superior rendering performance.
Authors: Antonio Emanuele Cin\`a, J\'er\^ome Rony, Maura Pintor, Luca Demetrio, Ambra Demontis, Battista Biggio, Ismail Ben Ayed, Fabio Roli
Abstract: Adversarial examples are typically optimized with gradient-based attacks. While novel attacks are continuously proposed, each is shown to outperform its predecessors using different experimental setups, hyperparameter settings, and number of forward and backward calls to the target models. This provides overly-optimistic and even biased evaluations that may unfairly favor one particular attack over the others. In this work, we aim to overcome these limitations by proposing AttackBench, i.e., the first evaluation framework that enables a fair comparison among different attacks. To this end, we first propose a categorization of gradient-based attacks, identifying their main components and differences. We then introduce our framework, which evaluates their effectiveness and efficiency. We measure these characteristics by (i) defining an optimality metric that quantifies how close an attack is to the optimal solution, and (ii) limiting the number of forward and backward queries to the model, such that all attacks are compared within a given maximum query budget. Our extensive experimental analysis compares more than 100 attack implementations with a total of over 800 different configurations against CIFAR-10 and ImageNet models, highlighting that only very few attacks outperform all the competing approaches. Within this analysis, we shed light on several implementation issues that prevent many attacks from finding better solutions or running at all. We release AttackBench as a publicly available benchmark, aiming to continuously update it to include and evaluate novel gradient-based attacks for optimizing adversarial examples.
Authors: Kristina Mach, Hessam Roodaki, Michael Sommersperger, Nassir Navab
Abstract: This paper presents an innovative approach to intraoperative Optical Coherence Tomography (iOCT) image segmentation in ophthalmic surgery, leveraging statistical analysis of speckle patterns to incorporate statistical pathology-specific prior knowledge. Our findings indicate statistically different speckle patterns within the retina and between retinal layers and surgical tools, facilitating the segmentation of previously unseen data without the necessity for manual labeling. The research involves fitting various statistical distributions to iOCT data, enabling the differentiation of different ocular structures and surgical tools. The proposed segmentation model aims to refine the statistical findings based on prior tissue understanding to leverage statistical and biological knowledge. Incorporating statistical parameters, physical analysis of light-tissue interaction, and deep learning informed by biological structures enhance segmentation accuracy, offering potential benefits to real-time applications in ophthalmic surgical procedures. The study demonstrates the adaptability and precision of using Gamma distribution parameters and the derived binary maps as sole inputs for segmentation, notably enhancing the model's inference performance on unseen data.
Authors: Konstantinos Pasvantis, Eftychios Protopapadakis
Abstract: The application of deep learning models in medical diagnosis has showcased considerable efficacy in recent years. Nevertheless, a notable limitation involves the inherent lack of explainability during decision-making processes. This study addresses such a constraint, by enhancing the interpretability robustness. The primary focus is directed towards refining the explanations generated by the LIME Library and LIME image explainer. This is achieved throuhg post-processing mechanisms, based on scenario-specific rules. Multiple experiments have been conducted using publicly accessible datasets related to brain tumor detection. Our proposed post-heuristic approach demonstrates significant advancements, yielding more robust and concrete results, in the context of medical diagnosis.
Authors: Andr\'es Bell-Navas, Nourelhouda Groun, Mar\'ia Villalba-Orero, Enrique Lara-Pezzi, Jes\'us Garicano-Mena, Soledad Le Clainche
Abstract: Heart diseases are the main international cause of human defunction. According to the WHO, nearly 18 million people decease each year because of heart diseases. Also considering the increase of medical data, much pressure is put on the health industry to develop systems for early and accurate heart disease recognition. In this work, an automatic cardiac pathology recognition system based on a novel deep learning framework is proposed, which analyses in real-time echocardiography video sequences. The system works in two stages. The first one transforms the data included in a database of echocardiography sequences into a machine-learning-compatible collection of annotated images which can be used in the training stage of any kind of machine learning-based framework, and more specifically with deep learning. This includes the use of the Higher Order Dynamic Mode Decomposition (HODMD) algorithm, for the first time to the authors' knowledge, for both data augmentation and feature extraction in the medical field. The second stage is focused on building and training a Vision Transformer (ViT), barely explored in the related literature. The ViT is adapted for an effective training from scratch, even with small datasets. The designed neural network analyses images from an echocardiography sequence to predict the heart state. The results obtained show the superiority of the proposed system and the efficacy of the HODMD algorithm, even outperforming pretrained Convolutional Neural Networks (CNNs), which are so far the method of choice in the literature.
Authors: Marwa Afnouch, Fares Bougourzi, Olfa Gaddour, Fadi Dornaika, Abdelmalik Taleb-Ahmed
Abstract: In recent years, Artificial Intelligence (AI) has been widely used in medicine, particularly in the analysis of medical imaging, which has been driven by advances in computer vision and deep learning methods. This is particularly important in overcoming the challenges posed by diseases such as Bone Metastases (BM), a common and complex malignancy of the bones. Indeed, there have been an increasing interest in developing Machine Learning (ML) techniques into oncologic imaging for BM analysis. In order to provide a comprehensive overview of the current state-of-the-art and advancements for BM analysis using artificial intelligence, this review is conducted with the accordance with PRISMA guidelines. Firstly, this review highlights the clinical and oncologic perspectives of BM and the used medical imaging modalities, with discussing their advantages and limitations. Then the review focuses on modern approaches with considering the main BM analysis tasks, which includes: classification, detection and segmentation. The results analysis show that ML technologies can achieve promising performance for BM analysis and have significant potential to improve clinician efficiency and cope with time and cost limitations. Furthermore, there are requirements for further research to validate the clinical performance of ML tools and facilitate their integration into routine clinical practice.
Authors: Emmanuelle Bourigault, Abdullah Hamdi, Amir Jamaludin
Abstract: In this work, we present X-Diffusion, a cross-sectional diffusion model tailored for Magnetic Resonance Imaging (MRI) data. X-Diffusion is capable of generating the entire MRI volume from just a single MRI slice or optionally from few multiple slices, setting new benchmarks in the precision of synthesized MRIs from extremely sparse observations. The uniqueness lies in the novel view-conditional training and inference of X-Diffusion on MRI volumes, allowing for generalized MRI learning. Our evaluations span both brain tumour MRIs from the BRATS dataset and full-body MRIs from the UK Biobank dataset. Utilizing the paired pre-registered Dual-energy X-ray Absorptiometry (DXA) and MRI modalities in the UK Biobank dataset, X-Diffusion is able to generate detailed 3D MRI volume from a single full-body DXA. Remarkably, the resultant MRIs not only stand out in precision on unseen examples (surpassing state-of-the-art results by large margins) but also flawlessly retain essential features of the original MRI, including tumour profiles, spine curvature, brain volume, and beyond. Furthermore, the trained X-Diffusion model on the MRI datasets attains a generalization capacity out-of-domain (e.g. generating knee MRIs even though it is trained on brains). The code is available on the project website https://emmanuelleb985.github.io/XDiffusion/ .
Authors: James Koch, Brenda Forland, Bruce Bernacki, Timothy Doster, Tegan Emerson
Abstract: We present a framework for inferring an atmospheric transmission profile from a spectral scene. This framework leverages a lightweight, physics-based simulator that is automatically tuned - by virtue of autodifferentiation and differentiable programming - to construct a surrogate atmospheric profile to model the observed data. We demonstrate utility of the methodology by (i) performing atmospheric correction, (ii) recasting spectral data between various modalities (e.g. radiance and reflectance at the surface and at the sensor), and (iii) inferring atmospheric transmission profiles, such as absorbing bands and their relative magnitudes.
Authors: Shivam Mehta, Anna Deichler, Jim O'Regan, Birger Mo\"ell, Jonas Beskow, Gustav Eje Henter, Simon Alexanderson
Abstract: Although humans engaged in face-to-face conversation simultaneously communicate both verbally and non-verbally, methods for joint and unified synthesis of speech audio and co-speech 3D gesture motion from text are a new and emerging field. These technologies hold great promise for more human-like, efficient, expressive, and robust synthetic communication, but are currently held back by the lack of suitably large datasets, as existing methods are trained on parallel data from all constituent modalities. Inspired by student-teacher methods, we propose a straightforward solution to the data shortage, by simply synthesising additional training material. Specifically, we use unimodal synthesis models trained on large datasets to create multimodal (but synthetic) parallel training data, and then pre-train a joint synthesis model on that material. In addition, we propose a new synthesis architecture that adds better and more controllable prosody modelling to the state-of-the-art method in the field. Our results confirm that pre-training on large amounts of synthetic data improves the quality of both the speech and the motion synthesised by the multimodal model, with the proposed architecture yielding further benefits when pre-trained on the synthetic data. See https://shivammehta25.github.io/MAGI/ for example output.
Authors: Yunzhen Feng, Tim G. J. Rudner, Nikolaos Tsilivis, Julia Kempe
Abstract: Adversarial examples have been shown to cause neural networks to fail on a wide range of vision and language tasks, but recent work has claimed that Bayesian neural networks (BNNs) are inherently robust to adversarial perturbations. In this work, we examine this claim. To study the adversarial robustness of BNNs, we investigate whether it is possible to successfully break state-of-the-art BNN inference methods and prediction pipelines using even relatively unsophisticated attacks for three tasks: (1) label prediction under the posterior predictive mean, (2) adversarial example detection with Bayesian predictive uncertainty, and (3) semantic shift detection. We find that BNNs trained with state-of-the-art approximate inference methods, and even BNNs trained with Hamiltonian Monte Carlo, are highly susceptible to adversarial attacks. We also identify various conceptual and experimental errors in previous works that claimed inherent adversarial robustness of BNNs and conclusively demonstrate that BNNs and uncertainty-aware Bayesian prediction pipelines are not inherently robust against adversarial attacks.
Authors: Ge Yan, Yaniv Romano, Tsui-Wei Weng
Abstract: Conformal prediction is a powerful tool to generate uncertainty sets with guaranteed coverage using any predictive model, under the assumption that the training and test data are i.i.d.. Recently, it has been shown that adversarial examples are able to manipulate conformal methods to construct prediction sets with invalid coverage rates, as the i.i.d. assumption is violated. To address this issue, a recent work, Randomized Smoothed Conformal Prediction (RSCP), was first proposed to certify the robustness of conformal prediction methods to adversarial noise. However, RSCP has two major limitations: (i) its robustness guarantee is flawed when used in practice and (ii) it tends to produce large uncertainty sets. To address these limitations, we first propose a novel framework called RSCP+ to provide provable robustness guarantee in evaluation, which fixes the issues in the original RSCP method. Next, we propose two novel methods, Post-Training Transformation (PTT) and Robust Conformal Training (RCT), to effectively reduce prediction set size with little computation overhead. Experimental results in CIFAR10, CIFAR100, and ImageNet suggest the baseline method only yields trivial predictions including full label set, while our methods could boost the efficiency by up to $4.36\times$, $5.46\times$, and $16.9\times$ respectively and provide practical robustness guarantee. Our codes are available at https://github.com/Trustworthy-ML-Lab/Provably-Robust-Conformal-Prediction.
URLs: https://github.com/Trustworthy-ML-Lab/Provably-Robust-Conformal-Prediction.
Authors: Yuto Nakashima, Mingzhe Yang, Yukino Baba
Abstract: Generating preferred images using generative adversarial networks (GANs) is challenging owing to the high-dimensional nature of latent space. In this study, we propose a novel approach that uses simple user-swipe interactions to generate preferred images for users. To effectively explore the latent space with only swipe interactions, we apply principal component analysis to the latent space of the StyleGAN, creating meaningful subspaces. We use a multi-armed bandit algorithm to decide the dimensions to explore, focusing on the preferences of the user. Experiments show that our method is more efficient in generating preferred images than the baseline methods. Furthermore, changes in preferred images during image generation or the display of entirely different image styles were observed to provide new inspirations, subsequently altering user preferences. This highlights the dynamic nature of user preferences, which our proposed approach recognizes and enhances.
Authors: Wei Yin, Yifan Liu, Chunhua Shen, Baichuan Sun, Anton van den Hengel
Abstract: We propose an approach to semantic segmentation that achieves state-of-the-art supervised performance when applied in a zero-shot setting. It thus achieves results equivalent to those of the supervised methods, on each of the major semantic segmentation datasets, without training on those datasets. This is achieved by replacing each class label with a vector-valued embedding of a short paragraph that describes the class. The generality and simplicity of this approach enables merging multiple datasets from different domains, each with varying class labels and semantics. The resulting merged semantic segmentation dataset of over 2 Million images enables training a model that achieves performance equal to that of state-of-the-art supervised methods on 7 benchmark datasets, despite not using any images therefrom. By fine-tuning the model on standard semantic segmentation datasets, we also achieve a significant improvement over the state-of-the-art supervised segmentation on NYUD-V2 and PASCAL-context at 60% and 65% mIoU, respectively. Based on the closeness of language embeddings, our method can even segment unseen labels. Extensive experiments demonstrate strong generalization to unseen image domains and unseen labels, and that the method enables impressive performance improvements in downstream applications, including depth estimation and instance segmentation.
Authors: Yidong Zhao, Changchun Yang, Artur Schweidtmann, Qian Tao
Abstract: The self-configuring nnU-Net has achieved leading performance in a large range of medical image segmentation challenges. It is widely considered as the model of choice and a strong baseline for medical image segmentation. However, despite its extraordinary performance, nnU-Net does not supply a measure of uncertainty to indicate its possible failure. This can be problematic for large-scale image segmentation applications, where data are heterogeneous and nnU-Net may fail without notice. In this work, we introduce a novel method to estimate nnU-Net uncertainty for medical image segmentation. We propose a highly effective scheme for posterior sampling of weight space for Bayesian uncertainty estimation. Different from previous baseline methods such as Monte Carlo Dropout and mean-field Bayesian Neural Networks, our proposed method does not require a variational architecture and keeps the original nnU-Net architecture intact, thereby preserving its excellent performance and ease of use. Additionally, we boost the segmentation performance over the original nnU-Net via marginalizing multi-modal posterior models. We applied our method on the public ACDC and M&M datasets of cardiac MRI and demonstrated improved uncertainty estimation over a range of baseline methods. The proposed method further strengthens nnU-Net for medical image segmentation in terms of both segmentation accuracy and quality control.
Authors: Zhihao Qian, Yutian Lin, Bo Du
Abstract: Visible-infrared person re-identification (VI-ReID) aims to retrieve images of the same pedestrian from different modalities, where the challenges lie in the significant modality discrepancy. To alleviate the modality gap, recent methods generate intermediate images by GANs, grayscaling, or mixup strategies. However, these methods could introduce extra data distribution, and the semantic correspondence between the two modalities is not well learned. In this paper, we propose a Patch-Mixed Cross-Modality framework (PMCM), where two images of the same person from two modalities are split into patches and stitched into a new one for model learning. A part-alignment loss is introduced to regularize representation learning, and a patch-mixed modality learning loss is proposed to align between the modalities. In this way, the model learns to recognize a person through patches of different styles, thereby the modality semantic correspondence can be inferred. In addition, with the flexible image generation strategy, the patch-mixed images freely adjust the ratio of different modality patches, which could further alleviate the modality imbalance problem. On two VI-ReID datasets, we report new state-of-the-art performance with the proposed method.
Authors: Arpit Garg, Cuong Nguyen, Rafael Felix, Thanh-Toan Do, Gustavo Carneiro
Abstract: The prevalence of noisy-label samples poses a significant challenge in deep learning, inducing overfitting effects. This has, therefore, motivated the emergence of learning with noisy-label (LNL) techniques that focus on separating noisy- and clean-label samples to apply different learning strategies to each group of samples. Current methodologies often rely on the small-loss hypothesis or feature-based selection to separate noisy- and clean-label samples, yet our empirical observations reveal their limitations, especially for labels with instance dependent noise (IDN). An important characteristic of IDN is the difficulty to distinguish the clean-label samples that lie near the decision boundary (i.e., the hard samples) from the noisy-label samples. We, therefore, propose a new noisy-label detection method, termed Peer-Agreement based Sample Selection (PASS), to address this problem. Utilising a trio of classifiers, PASS employs consensus-driven peer-based agreement of two models to select the samples to train the remaining model. PASS is easily integrated into existing LNL models, enabling the improvement of the detection accuracy of noisy- and clean-label samples, which increases the classification accuracy across various LNL benchmarks.
Authors: David Rozenberszki, Or Litany, Angela Dai
Abstract: 3D instance segmentation is fundamental to geometric understanding of the world around us. Existing methods for instance segmentation of 3D scenes rely on supervision from expensive, manual 3D annotations. We propose UnScene3D, the first fully unsupervised 3D learning approach for class-agnostic 3D instance segmentation of indoor scans. UnScene3D first generates pseudo masks by leveraging self-supervised color and geometry features to find potential object regions. We operate on a basis of geometric oversegmentation, enabling efficient representation and learning on high-resolution 3D data. The coarse proposals are then refined through self-training our model on its predictions. Our approach improves over state-of-the-art unsupervised 3D instance segmentation methods by more than 300% Average Precision score, demonstrating effective instance segmentation even in challenging, cluttered 3D scenes.
Authors: Ali Hojjat, Janek Haberer, Olaf Landsiedel
Abstract: Progressive compression allows images to start loading as low-resolution versions, becoming clearer as more data is received. This increases user experience when, for example, network connections are slow. Today, most approaches for image compression, both classical and learned ones, are designed to be non-progressive. This paper introduces ProgDTD, a training method that transforms learned, non-progressive image compression approaches into progressive ones. The design of ProgDTD is based on the observation that the information stored within the bottleneck of a compression model commonly varies in importance. To create a progressive compression model, ProgDTD modifies the training steps to enforce the model to store the data in the bottleneck sorted by priority. We achieve progressive compression by transmitting the data in order of its sorted index. ProgDTD is designed for CNN-based learned image compression models, does not need additional parameters, and has a customizable range of progressiveness. For evaluation, we apply ProgDTDto the hyperprior model, one of the most common structures in learned image compression. Our experimental results show that ProgDTD performs comparably to its non-progressive counterparts and other state-of-the-art progressive models in terms of MS-SSIM and accuracy.
Authors: Yingjie Tian, Yiqi Wang, Xianda Guo, Zheng Zhu, Long Chen
Abstract: In recent years, soft prompt learning methods have been proposed to fine-tune large-scale vision-language pre-trained models for various downstream tasks. These methods typically combine learnable textual tokens with class tokens as input for models with frozen parameters. However, they often employ a single prompt to describe class contexts, failing to capture categories' diverse attributes adequately. This study introduces the Partitioned Multi-modal Prompt (PMPO), a multi-modal prompting technique that extends the soft prompt from a single learnable prompt to multiple prompts. Our method divides the visual encoder depths and connects learnable prompts to the separated visual depths, enabling different prompts to capture the hierarchical contextual depths of visual representations. Furthermore, to maximize the advantages of multi-prompt learning, we incorporate prior information from manually designed templates and learnable multi-prompts, thus improving the generalization capabilities of our approach. We evaluate the effectiveness of our approach on three challenging tasks: new class generalization, cross-dataset evaluation, and domain generalization. For instance, our method achieves a $79.28$ harmonic mean, averaged over 11 diverse image recognition datasets ($+7.62$ compared to CoOp), demonstrating significant competitiveness compared to state-of-the-art prompting methods.
Authors: Longkang Peng, Tao Wei, Xuehong Chen, Xiaobei Chen, Rui Sun, Luoma Wan, Jin Chen, Xiaolin Zhu
Abstract: Convolutional neural networks (ConvNets) have been successfully applied to satellite image scene classification. Human-labeled training datasets are essential for ConvNets to perform accurate classification. Errors in human-annotated training datasets are unavoidable due to the complexity of satellite images. However, the distribution of real-world human-annotated label noises on remote sensing images and their impact on ConvNets have not been investigated. To fill this research gap, this study, for the first time, collected real-world labels from 32 participants and explored how their annotated label noise affect three representative ConvNets (VGG16, GoogleNet, and ResNet-50) for remote sensing image scene classification. We found that: (1) human-annotated label noise exhibits significant class and instance dependence; (2) an additional 1% of human-annotated label noise in training data leads to 0.5% reduction in the overall accuracy of ConvNets classification; (3) the error pattern of ConvNet predictions was strongly correlated with that of participant's labels. To uncover the mechanism underlying the impact of human labeling errors on ConvNets, we further compared it with three types of simulated label noise: uniform noise, class-dependent noise and instance-dependent noise. Our results show that the impact of human-annotated label noise on ConvNets significantly differs from all three types of simulated label noise, while both class dependence and instance dependence contribute to the impact of human-annotated label noise on ConvNets. These observations necessitate a reevaluation of the handling of noisy labels, and we anticipate that our real-world label noise dataset would facilitate the future development and assessment of label-noise learning algorithms.
Authors: Changyuan Wang, Ziwei Wang, Xiuwei Xu, Yansong Tang, Jie Zhou, Jiwen Lu
Abstract: In this paper, we propose an accurate data-free post-training quantization framework of diffusion models (ADP-DM) for efficient image generation. Conventional data-free quantization methods learn shared quantization functions for tensor discretization regardless of the generation timesteps, while the activation distribution differs significantly across various timesteps. The calibration images are acquired in random timesteps which fail to provide sufficient information for generalizable quantization function learning. Both issues cause sizable quantization errors with obvious image generation performance degradation. On the contrary, we design group-wise quantization functions for activation discretization in different timesteps and sample the optimal timestep for informative calibration image generation, so that our quantized diffusion model can reduce the discretization errors with negligible computational overhead. Specifically, we partition the timesteps according to the importance weights of quantization functions in different groups, which are optimized by differentiable search algorithms. We also select the optimal timestep for calibration image generation by structural risk minimizing principle in order to enhance the generalization ability in the deployment of quantized diffusion model. Extensive experimental results show that our method outperforms the state-of-the-art post-training quantization of diffusion model by a sizable margin with similar computational cost.
Authors: Arpit Garg, Cuong Nguyen, Rafael Felix, Thanh-Toan Do, Gustavo Carneiro
Abstract: Deep learning faces a formidable challenge when handling noisy labels, as models tend to overfit samples affected by label noise. This challenge is further compounded by the presence of instance-dependent noise (IDN), a realistic form of label noise arising from ambiguous sample information. To address IDN, Label Noise Learning (LNL) incorporates a sample selection stage to differentiate clean and noisy-label samples. This stage uses an arbitrary criterion and a pre-defined curriculum that initially selects most samples as noisy and gradually decreases this selection rate during training. Such curriculum is sub-optimal since it does not consider the actual label noise rate in the training set. This paper addresses this issue with a new noise-rate estimation method that is easily integrated with most state-of-the-art (SOTA) LNL methods to produce a more effective curriculum. Synthetic and real-world benchmark results demonstrate that integrating our approach with SOTA LNL methods improves accuracy in most cases.
Authors: Seunghyeok Back, Sangbeom Lee, Kangmin Kim, Joosoon Lee, Sungho Shin, Jemo Maeng, Kyoobin Lee
Abstract: Accurate perception of unknown objects is essential for autonomous robots, particularly when manipulating novel items in unstructured environments. However, existing unknown object instance segmentation (UOIS) methods often have over-segmentation and under-segmentation problems, resulting in inaccurate instance boundaries and failures in subsequent robotic tasks such as grasping and placement. To address this challenge, this article introduces INSTA-BEER, a fast and accurate model-agnostic refinement method that enhances the UOIS performance. The model adopts an error-informed refinement approach, which first predicts pixel-wise errors in the initial segmentation and then refines the segmentation guided by these error estimates. We introduce the quad-metric boundary error, which quantifies pixel-wise true positives, true negatives, false positives, and false negatives at the boundaries of object instances, effectively capturing both fine-grained and instance-level segmentation errors. Additionally, the Error Guidance Fusion (EGF) module explicitly integrates error information into the refinement process, further improving segmentation quality. In comprehensive evaluations conducted on three widely used benchmark datasets, INSTA-BEER outperformed state-of-the-art models in both accuracy and inference time. Moreover, a real-world robotic experiment demonstrated the practical applicability of our method in improving the performance of target object grasping tasks in cluttered environments.
Authors: Thomas Braure, Delphine Lazaro, David Hateau, Vincent Brandon, K\'evin Ginsburger
Abstract: Computed Tomography (CT) is a prominent example of Imaging Inverse Problem highlighting the unrivaled performances of data-driven methods in degraded measurements setups like sparse X-ray projections. Although a significant proportion of deep learning approaches benefit from large supervised datasets, they cannot generalize to new experimental setups. In contrast, fully unsupervised techniques, most notably using score-based generative models, have recently demonstrated similar or better performances compared to supervised approaches while being flexible at test time. However, their use cases are limited as they need considerable amounts of training data to have good generalization properties. Another unsupervised approach taking advantage of the implicit natural bias of deep convolutional networks, Deep Image Prior, has recently been adapted to solve sparse CT by reparameterizing the reconstruction problem. Although this methodology does not require any training dataset, it enforces a weaker prior on the reconstructions when compared to data-driven methods. To fill the gap between these two strategies, we propose an unsupervised conditional approach to the Generative Latent Optimization framework (cGLO). Similarly to DIP, without any training dataset, cGLO benefits from the structural bias of a decoder network. However, the prior is further reinforced as the effect of a likelihood objective shared between multiple slices being reconstructed simultaneously through the same decoder network. In addition, the parameters of the decoder may be initialized on an unsupervised, and eventually very small, training dataset to enhance the reconstruction. The resulting approach is tested on full-dose sparse-view CT using multiple training dataset sizes and varying numbers of viewing angles.
Authors: Songtao Luo, Shuang Yang, Shiguang Shan, Xilin Chen
Abstract: In this paper, we propose a novel method for speaker adaptation in lip reading, motivated by two observations. Firstly, a speaker's own characteristics can always be portrayed well by his/her few facial images or even a single image with shallow networks, while the fine-grained dynamic features associated with speech content expressed by the talking face always need deep sequential networks to represent accurately. Therefore, we treat the shallow and deep layers differently for speaker adaptive lip reading. Secondly, we observe that a speaker's unique characteristics ( e.g. prominent oral cavity and mandible) have varied effects on lip reading performance for different words and pronunciations, necessitating adaptive enhancement or suppression of the features for robust lip reading. Based on these two observations, we propose to take advantage of the speaker's own characteristics to automatically learn separable hidden unit contributions with different targets for shallow layers and deep layers respectively. For shallow layers where features related to the speaker's characteristics are stronger than the speech content related features, we introduce speaker-adaptive features to learn for enhancing the speech content features. For deep layers where both the speaker's features and the speech content features are all expressed well, we introduce the speaker-adaptive features to learn for suppressing the speech content irrelevant noise for robust lip reading. Our approach consistently outperforms existing methods, as confirmed by comprehensive analysis and comparison across different settings. Besides the evaluation on the popular LRW-ID and GRID datasets, we also release a new dataset for evaluation, CAS-VSR-S68h, to further assess the performance in an extreme setting where just a few speakers are available but the speech content covers a large and diversified range.
Authors: Che Liu, Anand Shah, Wenjia Bai, Rossella Arcucci
Abstract: Medical Vision-Language Pre-training (VLP) learns representations jointly from medical images and paired radiology reports. It typically requires large-scale paired image-text datasets to achieve effective pre-training for both the image encoder and text encoder. The advent of text-guided generative models raises a compelling question: Can VLP be implemented solely with synthetic images generated from genuine radiology reports, thereby mitigating the need for extensively pairing and curating image-text datasets? In this work, we scrutinize this very question by examining the feasibility and effectiveness of employing synthetic images for medical VLP. We replace real medical images with their synthetic equivalents, generated from authentic medical reports. Utilizing three state-of-the-art VLP algorithms, we exclusively train on these synthetic samples. Our empirical evaluation across three subsequent tasks, namely image classification, semantic segmentation and object detection, reveals that the performance achieved through synthetic data is on par with or even exceeds that obtained with real images. As a pioneering contribution to this domain, we introduce a large-scale synthetic medical image dataset, paired with anonymized real radiology reports. This alleviates the need of sharing medical images, which are not easy to curate and share in practice. The code and the dataset can be found in \href{https://github.com/cheliu-computation/MedSyn-RepLearn/tree/main}{https://github.com/cheliu-computation/MedSyn-RepLearn/tree/main}.
URLs: https://github.com/cheliu-computation/MedSyn-RepLearn/tree/main, https://github.com/cheliu-computation/MedSyn-RepLearn/tree/main
Authors: Che Liu, Sibo Cheng, Miaojing Shi, Anand Shah, Wenjia Bai, Rossella Arcucci
Abstract: In the field of medical Vision-Language Pre-training (VLP), significant efforts have been devoted to deriving text and image features from both clinical reports and associated medical images. However, most existing methods may have overlooked the opportunity in leveraging the inherent hierarchical structure of clinical reports, which are generally split into `findings' for descriptive content and `impressions' for conclusive observation. Instead of utilizing this rich, structured format, current medical VLP approaches often simplify the report into either a unified entity or fragmented tokens. In this work, we propose a novel clinical prior guided VLP framework named IMITATE to learn the structure information from medical reports with hierarchical vision-language alignment. The framework derives multi-level visual features from the chest X-ray (CXR) images and separately aligns these features with the descriptive and the conclusive text encoded in the hierarchical medical report. Furthermore, a new clinical-informed contrastive loss is introduced for cross-modal learning, which accounts for clinical prior knowledge in formulating sample correlations in contrastive learning. The proposed model, IMITATE, outperforms baseline VLP methods across six different datasets, spanning five medical imaging downstream tasks. Comprehensive experimental results highlight the advantages of integrating the hierarchical structure of medical reports for vision-language alignment.
Authors: Lorenzo Luzi, Helen Jenne, Ryan Murray, Carlos Ortiz Marrero
Abstract: The rapid advancement of Generative Adversarial Networks (GANs) necessitates the need to robustly evaluate these models. Among the established evaluation criteria, the Fr\'{e}chetInception Distance (FID) has been widely adopted due to its conceptual simplicity, fast computation time, and strong correlation with human perception. However, FID has inherent limitations, mainly stemming from its assumption that feature embeddings follow a Gaussian distribution, and therefore can be defined by their first two moments. As this does not hold in practice, in this paper we explore the importance of third-moments in image feature data and use this information to define a new measure, which we call the Skew Inception Distance (SID). We prove that SID is a pseudometric on probability distributions, show how it extends FID, and present a practical method for its computation. Our numerical experiments support that SID either tracks with FID or, in some cases, aligns more closely with human perception when evaluating image features of ImageNet data. Our work also shows that principal component analysis can be used to speed up the computation time of both FID and SID. Although we focus on using SID on image features for GAN evaluation, SID is applicable much more generally, including for the evaluation of other generative models.
Authors: Sungjune Park, Hyunjun Kim, Yong Man Ro
Abstract: Large language models (LLMs) have shown their capabilities in understanding contextual and semantic information regarding knowledge of instance appearances. In this paper, we introduce a novel approach to utilize the strengths of LLMs in understanding contextual appearance variations and to leverage this knowledge into a vision model (here, pedestrian detection). While pedestrian detection is considered one of the crucial tasks directly related to our safety (e.g., intelligent driving systems), it is challenging because of varying appearances and poses in diverse scenes. Therefore, we propose to formulate language-derived appearance elements and incorporate them with visual cues in pedestrian detection. To this end, we establish a description corpus that includes numerous narratives describing various appearances of pedestrians and other instances. By feeding them through an LLM, we extract appearance knowledge sets that contain the representations of appearance variations. Subsequently, we perform a task-prompting process to obtain appearance elements which are guided representative appearance knowledge relevant to a downstream pedestrian detection task. The obtained knowledge elements are adaptable to various detection frameworks, so that we can provide plentiful appearance information by integrating the language-derived appearance elements with visual cues within a detector. Through comprehensive experiments with various pedestrian detectors, we verify the adaptability and effectiveness of our method showing noticeable performance gains and achieving state-of-the-art detection performance on two public pedestrian detection benchmarks (i.e., CrowdHuman and WiderPedestrian).
Authors: Luk\'a\v{s} Adam, Vojt\v{e}ch \v{C}erm\'ak, Kostas Papafitsoros, Luk\'a\v{s} Picek
Abstract: This paper introduces the first public large-scale, long-span dataset with sea turtle photographs captured in the wild -- SeaTurtleID2022 (https://www.kaggle.com/datasets/wildlifedatasets/seaturtleid2022). The dataset contains 8729 photographs of 438 unique individuals collected within 13 years, making it the longest-spanned dataset for animal re-identification. All photographs include various annotations, e.g., identity, encounter timestamp, and body parts segmentation masks. Instead of standard "random" splits, the dataset allows for two realistic and ecologically motivated splits: (i) a time-aware closed-set with training, validation, and test data from different days/years, and (ii) a time-aware open-set with new unknown individuals in test and validation sets. We show that time-aware splits are essential for benchmarking re-identification methods, as random splits lead to performance overestimation. Furthermore, a baseline instance segmentation and re-identification performance over various body parts is provided. Finally, an end-to-end system for sea turtle re-identification is proposed and evaluated. The proposed system based on Hybrid Task Cascade for head instance segmentation and ArcFace-trained feature-extractor achieved an accuracy of 86.8%.
URLs: https://www.kaggle.com/datasets/wildlifedatasets/seaturtleid2022).
Authors: Feifei Wang, Zhentao Tan, Tianyi Wei, Yue Wu, Qidong Huang
Abstract: Despite the success of diffusion-based customization methods on visual content creation, increasing concerns have been raised about such techniques from both privacy and political perspectives. To tackle this issue, several anti-customization methods have been proposed in very recent months, predominantly grounded in adversarial attacks. Unfortunately, most of these methods adopt straightforward designs, such as end-to-end optimization with a focus on adversarially maximizing the original training loss, thereby neglecting nuanced internal properties intrinsic to the diffusion model, and even leading to ineffective optimization in some diffusion time steps.In this paper, we strive to bridge this gap by undertaking a comprehensive exploration of these inherent properties, to boost the performance of current anti-customization approaches. Two aspects of properties are investigated: 1) We examine the relationship between time step selection and the model's perception in the frequency domain of images and find that lower time steps can give much more contributions to adversarial noises. This inspires us to propose an adaptive greedy search for optimal time steps that seamlessly integrates with existing anti-customization methods. 2) We scrutinize the roles of features at different layers during denoising and devise a sophisticated feature-based optimization framework for anti-customization.Experiments on facial benchmarks demonstrate that our approach significantly increases identity disruption, thereby protecting user privacy and copyright. Our code is available at: https://github.com/somuchtome/SimAC.
Authors: Li Jiang, Zhaowei Lu
Abstract: Image forensics has become increasingly crucial in our daily lives. Among various types of forgeries, copy-move forgery detection has received considerable attention within the academic community. Keypoint-based algorithms, particularly those based on Scale Invariant Feature Transform, have achieved promising outcomes. However, most of keypoint detection algorithms failed to generate sufficient matches when tampered patches were occurred in smooth areas, leading to insufficient matches. Therefore, this paper introduces entropy images to determine the coordinates and scales of keypoints based on Scale Invariant Feature Transform detector, which make the pre-processing more suitable for solving the above problems. Furthermore, an overlapped entropy level clustering algorithm is developed to mitigate the increased matching complexity caused by the non-ideal distribution of gray values in keypoints. Experimental results demonstrate that our algorithm achieves a good balance between performance and time efficiency.
Authors: Zikai Zhou, Yunhang Shen, Shitong Shao, Linrui Gong, Shaohui Lin
Abstract: Knowledge distillation has emerged as a highly effective method for bridging the representation discrepancy between large-scale models and lightweight models. Prevalent approaches involve leveraging appropriate metrics to minimize the divergence or distance between the knowledge extracted from the teacher model and the knowledge learned by the student model. Centered Kernel Alignment (CKA) is widely used to measure representation similarity and has been applied in several knowledge distillation methods. However, these methods are complex and fail to uncover the essence of CKA, thus not answering the question of how to use CKA to achieve simple and effective distillation properly. This paper first provides a theoretical perspective to illustrate the effectiveness of CKA, which decouples CKA to the upper bound of Maximum Mean Discrepancy~(MMD) and a constant term. Drawing from this, we propose a novel Relation-Centered Kernel Alignment~(RCKA) framework, which practically establishes a connection between CKA and MMD. Furthermore, we dynamically customize the application of CKA based on the characteristics of each task, with less computational source yet comparable performance than the previous methods. The extensive experiments on the CIFAR-100, ImageNet-1k, and MS-COCO demonstrate that our method achieves state-of-the-art performance on almost all teacher-student pairs for image classification and object detection, validating the effectiveness of our approaches. Our code is available in https://github.com/Klayand/PCKA
Authors: Mike Laszkiewicz, Imant Daunhawer, Julia E. Vogt, Asja Fischer, Johannes Lederer
Abstract: Recent years have witnessed a rapid development of deep generative models for creating synthetic media, such as images and videos. While the practical applications of these models in everyday tasks are enticing, it is crucial to assess the inherent risks regarding their fairness. In this work, we introduce a comprehensive framework for benchmarking the performance and fairness of conditional generative models. We develop a set of metrics$\unicode{x2013}$inspired by their supervised fairness counterparts$\unicode{x2013}$to evaluate the models on their fairness and diversity. Focusing on the specific application of image upsampling, we create a benchmark covering a wide variety of modern upsampling methods. As part of the benchmark, we introduce UnfairFace, a subset of FairFace that replicates the racial distribution of common large-scale face datasets. Our empirical study highlights the importance of using an unbiased training set and reveals variations in how the algorithms respond to dataset imbalances. Alarmingly, we find that none of the considered methods produces statistically fair and diverse results. All experiments can be reproduced using our provided repository.
Authors: Shuai Yuan, Hanlin Qin, Xiang Yan, Naveed AKhtar, Ajmal Mian
Abstract: Infrared small target detection (IRSTD) has recently benefitted greatly from U-shaped neural models. However, largely overlooking effective global information modeling, existing techniques struggle when the target has high similarities with the background. We present a Spatial-channel Cross Transformer Network (SCTransNet) that leverages spatial-channel cross transformer blocks (SCTBs) on top of long-range skip connections to address the aforementioned challenge. In the proposed SCTBs, the outputs of all encoders are interacted with cross transformer to generate mixed features, which are redistributed to all decoders to effectively reinforce semantic differences between the target and clutter at full scales. Specifically, SCTB contains the following two key elements: (a) spatial-embedded single-head channel-cross attention (SSCA) for exchanging local spatial features and full-level global channel information to eliminate ambiguity among the encoders and facilitate high-level semantic associations of the images, and (b) a complementary feed-forward network (CFN) for enhancing the feature discriminability via a multi-scale strategy and cross-spatial-channel information interaction to promote beneficial information transfer. Our SCTransNet effectively encodes the semantic differences between targets and backgrounds to boost its internal representation for detecting small infrared targets accurately. Extensive experiments on three public datasets, NUDT-SIRST, NUAA-SIRST, and IRSTD-1k, demonstrate that the proposed SCTransNet outperforms existing IRSTD methods. Our code will be made public at https://github.com/xdFai.
Authors: Evan D. Cook, Marc-Antoine Lavoie, Steven L. Waslander
Abstract: Out-of-distribution (OOD) detection is a critical task for safe deployment of learning systems in the open world setting. In this work, we investigate the use of feature density estimation via normalizing flows for OOD detection and present a fully unsupervised approach which requires no exposure to OOD data, avoiding researcher bias in OOD sample selection. This is a post-hoc method which can be applied to any pretrained model, and involves training a lightweight auxiliary normalizing flow model to perform the out-of-distribution detection via density thresholding. Experiments on OOD detection in image classification show strong results for far-OOD data detection with only a single epoch of flow training, including 98.2% AUROC for ImageNet-1k vs. Textures, which exceeds the state of the art by 7.8%. We additionally explore the connection between the feature space distribution of the pretrained model and the performance of our method. Finally, we provide insights into training pitfalls that have plagued normalizing flows for use in OOD detection.
Authors: Chang Won Lee, Steven L. Waslander
Abstract: Multi-object tracking (MOT) methods have seen a significant boost in performance recently, due to strong interest from the research community and steadily improving object detection methods. The majority of tracking methods follow the tracking-by-detection (TBD) paradigm, blindly trust the incoming detections with no sense of their associated localization uncertainty. This lack of uncertainty awareness poses a problem in safety-critical tasks such as autonomous driving where passengers could be put at risk due to erroneous detections that have propagated to downstream tasks, including MOT. While there are existing works in probabilistic object detection that predict the localization uncertainty around the boxes, no work in 2D MOT for autonomous driving has studied whether these estimates are meaningful enough to be leveraged effectively in object tracking. We introduce UncertaintyTrack, a collection of extensions that can be applied to multiple TBD trackers to account for localization uncertainty estimates from probabilistic object detectors. Experiments on the Berkeley Deep Drive MOT dataset show that the combination of our method and informative uncertainty estimates reduces the number of ID switches by around 19\% and improves mMOTA by 2-3%. The source code is available at https://github.com/TRAILab/UncertaintyTrack
Authors: Shitao Tang, Jiacheng Chen, Dilin Wang, Chengzhou Tang, Fuyang Zhang, Yuchen Fan, Vikas Chandra, Yasutaka Furukawa, Rakesh Ranjan
Abstract: This paper presents a neural architecture MVDiffusion++ for 3D object reconstruction that synthesizes dense and high-resolution views of an object given one or a few images without camera poses. MVDiffusion++ achieves superior flexibility and scalability with two surprisingly simple ideas: 1) A ``pose-free architecture'' where standard self-attention among 2D latent features learns 3D consistency across an arbitrary number of conditional and generation views without explicitly using camera pose information; and 2) A ``view dropout strategy'' that discards a substantial number of output views during training, which reduces the training-time memory footprint and enables dense and high-resolution view synthesis at test time. We use the Objaverse for training and the Google Scanned Objects for evaluation with standard novel view synthesis and 3D reconstruction metrics, where MVDiffusion++ significantly outperforms the current state of the arts. We also demonstrate a text-to-3D application example by combining MVDiffusion++ with a text-to-image generative model. The project page is at https://mvdiffusion-plusplus.github.io.
Authors: Md Mohaiminul Islam, Ngan Ho, Xitong Yang, Tushar Nagarajan, Lorenzo Torresani, Gedas Bertasius
Abstract: Most video captioning models are designed to process short video clips of few seconds and output text describing low-level visual concepts (e.g., objects, scenes, atomic actions). However, most real-world videos last for minutes or hours and have a complex hierarchical structure spanning different temporal granularities. We propose Video ReCap, a recursive video captioning model that can process video inputs of dramatically different lengths (from 1 second to 2 hours) and output video captions at multiple hierarchy levels. The recursive video-language architecture exploits the synergy between different video hierarchies and can process hour-long videos efficiently. We utilize a curriculum learning training scheme to learn the hierarchical structure of videos, starting from clip-level captions describing atomic actions, then focusing on segment-level descriptions, and concluding with generating summaries for hour-long videos. Furthermore, we introduce Ego4D-HCap dataset by augmenting Ego4D with 8,267 manually collected long-range video summaries. Our recursive model can flexibly generate captions at different hierarchy levels while also being useful for other complex video understanding tasks, such as VideoQA on EgoSchema. Data, code, and models are available at: https://sites.google.com/view/vidrecap
Authors: Qixuan Zheng, Ming Zhang, Hong Yan
Abstract: To achieve greater accuracy, hypergraph matching algorithms require exponential increases in computational resources. Recent kd-tree-based approximate nearest neighbor (ANN) methods, despite the sparsity of their compatibility tensor, still require exhaustive calculations for large-scale graph matching. This work utilizes CUR tensor decomposition and introduces a novel cascaded second and third-order hypergraph matching framework (CURSOR) for efficient hypergraph matching. A CUR-based second-order graph matching algorithm is used to provide a rough match, and then the core of CURSOR, a fiber-CUR-based tensor generation method, directly calculates entries of the compatibility tensor by leveraging the initial second-order match result. This significantly decreases the time complexity and tensor density. A probability relaxation labeling (PRL)-based matching algorithm, especially suitable for sparse tensors, is developed. Experiment results on large-scale synthetic datasets and widely-adopted benchmark sets demonstrate the superiority of CURSOR over existing methods. The tensor generation method in CURSOR can be integrated seamlessly into existing hypergraph matching methods to improve their performance and lower their computational costs.
Authors: Tyler L. Hayes, C\'esar R. de Souza, Namil Kim, Jiwon Kim, Riccardo Volpi, Diane Larlus
Abstract: Object detectors are typically trained once and for all on a fixed set of classes. However, this closed-world assumption is unrealistic in practice, as new classes will inevitably emerge after the detector is deployed in the wild. In this work, we look at ways to extend a detector trained for a set of base classes so it can i) spot the presence of novel classes, and ii) automatically enrich its repertoire to be able to detect those newly discovered classes together with the base ones. We propose PANDAS, a method for novel class discovery and detection. It discovers clusters representing novel classes from unlabeled data, and represents old and new classes with prototypes. During inference, a distance-based classifier uses these prototypes to assign a label to each detected object instance. The simplicity of our method makes it widely applicable. We experimentally demonstrate the effectiveness of PANDAS on the VOC 2012 and COCO-to-LVIS benchmarks. It performs favorably against the state of the art for this task while being computationally more affordable.
Authors: Bo Yang, Hengwei Zhang, Jindong Wang, Yulong Yang, Chenhao Lin, Chao Shen, Zhengyu Zhao
Abstract: Transferable adversarial examples cause practical security risks since they can mislead a target model without knowing its internal knowledge. A conventional recipe for maximizing transferability is to keep only the optimal adversarial example from all those obtained in the optimization pipeline. In this paper, for the first time, we question this convention and demonstrate that those discarded, sub-optimal adversarial examples can be reused to boost transferability. Specifically, we propose ``Adversarial Example Soups'' (AES), with AES-tune for averaging discarded adversarial examples in hyperparameter tuning and AES-rand for stability testing. In addition, our AES is inspired by ``model soups'', which averages weights of multiple fine-tuned models for improved accuracy without increasing inference time. Extensive experiments validate the global effectiveness of our AES, boosting 10 state-of-the-art transfer attacks and their combinations by up to 13% against 10 diverse (defensive) target models. We also show the possibility of generalizing AES to other types, e.g., directly averaging multiple in-the-wild adversarial examples that yield comparable success. A promising byproduct of AES is the improved stealthiness of adversarial examples since the perturbation variances are naturally reduced.
Authors: Junzhe Zhang, Huixuan Zhang, Xunjian Yin, Xiaojun Wan
Abstract: News image captioning requires model to generate an informative caption rich in entities, with the news image and the associated news article. Though Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in addressing various vision-language tasks, our research finds that current MLLMs still bear limitations in handling entity information on news image captioning task. Besides, while MLLMs have the ability to process long inputs, generating high-quality news image captions still requires a trade-off between sufficiency and conciseness of textual input information. To explore the potential of MLLMs and address problems we discovered, we propose : an Entity-Aware Multimodal Alignment based approach for news image captioning. Our approach first aligns the MLLM through Balance Training Strategy with two extra alignment tasks: Entity-Aware Sentence Selection task and Entity Selection task, together with News Image Captioning task, to enhance its capability in handling multimodal entity information. The aligned MLLM will utilizes the additional entity-related information it explicitly extracts to supplement its textual input while generating news image captions. Our approach achieves better results than all previous models in CIDEr score on GoodNews dataset (72.33 -> 88.39) and NYTimes800k dataset (70.83 -> 85.61).
Authors: Zhiying Leng, Tolga Birdal, Xiaohui Liang, Federico Tombari
Abstract: 3D shape generation from text is a fundamental task in 3D representation learning. The text-shape pairs exhibit a hierarchical structure, where a general text like ``chair" covers all 3D shapes of the chair, while more detailed prompts refer to more specific shapes. Furthermore, both text and 3D shapes are inherently hierarchical structures. However, existing Text2Shape methods, such as SDFusion, do not exploit that. In this work, we propose HyperSDFusion, a dual-branch diffusion model that generates 3D shapes from a given text. Since hyperbolic space is suitable for handling hierarchical data, we propose to learn the hierarchical representations of text and 3D shapes in hyperbolic space. First, we introduce a hyperbolic text-image encoder to learn the sequential and multi-modal hierarchical features of text in hyperbolic space. In addition, we design a hyperbolic text-graph convolution module to learn the hierarchical features of text in hyperbolic space. In order to fully utilize these text features, we introduce a dual-branch structure to embed text features in 3D feature space. At last, to endow the generated 3D shapes with a hierarchical structure, we devise a hyperbolic hierarchical loss. Our method is the first to explore the hyperbolic hierarchical representation for text-to-shape generation. Experimental results on the existing text-to-shape paired dataset, Text2Shape, achieved state-of-the-art results. We release our implementation under HyperSDFusion.github.io.
Authors: Doanh C. Bui, Thinh V. Le, Ba Hung Ngo, Tae Jong Choi
Abstract: Person attribute recognition and attribute-based retrieval are two core human-centric tasks. In the recognition task, the challenge is specifying attributes depending on a person's appearance, while the retrieval task involves searching for matching persons based on attribute queries. There is a significant relationship between recognition and retrieval tasks. In this study, we demonstrate that if there is a sufficiently robust network to solve person attribute recognition, it can be adapted to facilitate better performance for the retrieval task. Another issue that needs addressing in the retrieval task is the modality gap between attribute queries and persons' images. Therefore, in this paper, we present CLEAR, a unified network designed to address both tasks. We introduce a robust cross-transformers network to handle person attribute recognition. Additionally, leveraging a pre-trained language model, we construct pseudo-descriptions for attribute queries and introduce an effective training strategy to train only a few additional parameters for adapters, facilitating the handling of the retrieval task. Finally, the unified CLEAR model is evaluated on five benchmarks: PETA, PA100K, Market-1501, RAPv2, and UPAR-2024. Without bells and whistles, CLEAR achieves state-of-the-art performance or competitive results for both tasks, significantly outperforming other competitors in terms of person retrieval performance on the widely-used Market-1501 dataset.
Authors: Anton Pelykh, Ozge Mercanoglu Sincan, Richard Bowden
Abstract: Recent years have seen significant progress in human image generation, particularly with the advancements in diffusion models. However, existing diffusion methods encounter challenges when producing consistent hand anatomy and the generated images often lack precise control over the hand pose. To address this limitation, we introduce a novel approach to pose-conditioned human image generation, dividing the process into two stages: hand generation and subsequent body outpainting around the hands. We propose training the hand generator in a multi-task setting to produce both hand images and their corresponding segmentation masks, and employ the trained model in the first stage of generation. An adapted ControlNet model is then used in the second stage to outpaint the body around the generated hands, producing the final result. A novel blending technique is introduced to preserve the hand details during the second stage that combines the results of both stages in a coherent way. This involves sequential expansion of the outpainted region while fusing the latent representations, to ensure a seamless and cohesive synthesis of the final image. Experimental evaluations demonstrate the superiority of our proposed method over state-of-the-art techniques, in both pose accuracy and image quality, as validated on the HaGRID dataset. Our approach not only enhances the quality of the generated hands but also offers improved control over hand pose, advancing the capabilities of pose-conditioned human image generation. The source code of the proposed approach is available at https://github.com/apelykh/hand-to-diffusion.
Authors: Xu Wu, XianXu Hou, Zhihui Lai, Jie Zhou, Ya-nan Zhang, Witold Pedrycz, Linlin Shen
Abstract: Low-light image enhancement (LLIE) aims to improve low-illumination images. However, existing methods face two challenges: (1) uncertainty in restoration from diverse brightness degradations; (2) loss of texture and color information caused by noise suppression and light enhancement. In this paper, we propose a novel enhancement approach, CodeEnhance, by leveraging quantized priors and image refinement to address these challenges. In particular, we reframe LLIE as learning an image-to-code mapping from low-light images to discrete codebook, which has been learned from high-quality images. To enhance this process, a Semantic Embedding Module (SEM) is introduced to integrate semantic information with low-level features, and a Codebook Shift (CS) mechanism, designed to adapt the pre-learned codebook to better suit the distinct characteristics of our low-light dataset. Additionally, we present an Interactive Feature Transformation (IFT) module to refine texture and color information during image reconstruction, allowing for interactive enhancement based on user preferences. Extensive experiments on both real-world and synthetic benchmarks demonstrate that the incorporation of prior knowledge and controllable information transfer significantly enhances LLIE performance in terms of quality and fidelity. The proposed CodeEnhance exhibits superior robustness to various degradations, including uneven illumination, noise, and color distortion.
Authors: He Wang, Pengcheng Guo, Xucheng Wan, Huan Zhou, Lei Xie
Abstract: Automatic lip-reading (ALR) aims to automatically transcribe spoken content from a speaker's silent lip motion captured in video. Current mainstream lip-reading approaches only use a single visual encoder to model input videos of a single scale. In this paper, we propose to enhance lip-reading by incorporating multi-scale video data and multi-encoder. Specifically, we first propose a novel multi-scale lip motion extraction algorithm based on the size of the speaker's face and an Enhanced ResNet3D visual front-end (VFE) to extract lip features at different scales. For the multi-encoder, in addition to the mainstream Transformer and Conformer, we also incorporate the recently proposed Branchformer and E-Branchformer as visual encoders. In the experiments, we explore the influence of different video data scales and encoders on ALR system performance and fuse the texts transcribed by all ALR systems using recognizer output voting error reduction (ROVER). Finally, our proposed approach placed second in the ICME 2024 ChatCLR Challenge Task 2, with a 21.52% reduction in character error rate (CER) compared to the official baseline on the evaluation set.
Authors: Sidra Aleem, Fangyijie Wang, Mayug Maniparambil, Eric Arazo, Julia Dietlmeier, Guenole Silvestre, Kathleen Curran, Noel E. O'Connor, Suzanne Little
Abstract: The Segment Anything Model (SAM) and CLIP are remarkable vision foundation models (VFMs). SAM, a prompt driven segmentation model, excels in segmentation tasks across diverse domains, while CLIP is renowned for its zero shot recognition capabilities. However, their unified potential has not yet been explored in medical image segmentation. To adapt SAM to medical imaging, existing methods primarily rely on tuning strategies that require extensive data or prior prompts tailored to the specific task, making it particularly challenging when only a limited number of data samples are available. This work presents an in depth exploration of integrating SAM and CLIP into a unified framework for medical image segmentation. Specifically, we propose a simple unified framework, SaLIP, for organ segmentation. Initially, SAM is used for part based segmentation within the image, followed by CLIP to retrieve the mask corresponding to the region of interest (ROI) from the pool of SAM generated masks. Finally, SAM is prompted by the retrieved ROI to segment a specific organ. Thus, SaLIP is training and fine tuning free and does not rely on domain expertise or labeled data for prompt engineering. Our method shows substantial enhancements in zero shot segmentation, showcasing notable improvements in DICE scores across diverse segmentation tasks like brain (63.46%), lung (50.11%), and fetal head (30.82%), when compared to un prompted SAM. Code and text prompts are available at: https://github.com/aleemsidra/SaLIP.
Authors: Xingyu Song, Zhan Li, Shi Chen, Xin-Qiang Cai, Kazuyuki Demachi
Abstract: Action recognition, an essential component of computer vision, plays a pivotal role in multiple applications. Despite significant improvements brought by Convolutional Neural Networks (CNNs), these models suffer performance declines when trained with discontinuous video frames, which is a frequent scenario in real-world settings. This decline primarily results from the loss of temporal continuity, which is crucial for understanding the semantics of human actions. To overcome this issue, we introduce the 4A (Action Animation-based Augmentation Approach) pipeline, which employs a series of sophisticated techniques: starting with 2D human pose estimation from RGB videos, followed by Quaternion-based Graph Convolution Network for joint orientation and trajectory prediction, and Dynamic Skeletal Interpolation for creating smoother, diversified actions using game engine technology. This innovative approach generates realistic animations in varied game environments, viewed from multiple viewpoints. In this way, our method effectively bridges the domain gap between virtual and real-world data. In experimental evaluations, the 4A pipeline achieves comparable or even superior performance to traditional training approaches using real-world data, while requiring only 10% of the original data volume. Additionally, our approach demonstrates enhanced performance on In-the-wild videos, marking a significant advancement in the field of action recognition.
Authors: Chunli Peng, Xuan Dong, Tiantian Cao, Zhengqing Li, Kun Dong, Weixin Li
Abstract: The fusion of images from dual camera systems featuring a wide-angle and a telephoto camera has become a hotspot problem recently. By integrating simultaneously captured wide-angle and telephoto images from these systems, the resulting fused image achieves a wide field of view (FOV) coupled with high-definition quality. Existing approaches are mostly deep learning methods, and predominantly rely on supervised learning, where the training dataset plays a pivotal role. However, current datasets typically adopt a data synthesis approach generate input pairs of wide-angle and telephoto images alongside ground-truth images. Notably, the wide-angle inputs are synthesized rather than captured using real wide-angle cameras, and the ground-truth image is captured by wide-angle camera whose quality is substantially lower than that of input telephoto images captured by telephoto cameras. To address these limitations, we introduce a novel hardware setup utilizing a beam splitter to simultaneously capture three images, i.e. input pairs and ground-truth images, from two authentic cellphones equipped with wide-angle and telephoto dual cameras. Specifically, the wide-angle and telephoto images captured by cellphone 2 serve as the input pair, while the telephoto image captured by cellphone 1, which is calibrated to match the optical path of the wide-angle image from cellphone 2, serves as the ground-truth image, maintaining quality on par with the input telephoto image. Experiments validate the efficacy of our newly introduced dataset, named ReWiTe, significantly enhances the performance of various existing methods for real-world wide-angle and telephoto dual image fusion tasks.
Authors: Zhong Wang, Zengyu Wan, Han Han, Bohao Liao, Yuliang Wu, Wei Zhai, Yang Cao, Zheng-jun Zha
Abstract: Event-based eye tracking has shown great promise with the high temporal resolution and low redundancy provided by the event camera. However, the diversity and abruptness of eye movement patterns, including blinking, fixating, saccades, and smooth pursuit, pose significant challenges for eye localization. To achieve a stable event-based eye-tracking system, this paper proposes a bidirectional long-term sequence modeling and time-varying state selection mechanism to fully utilize contextual temporal information in response to the variability of eye movements. Specifically, the MambaPupil network is proposed, which consists of the multi-layer convolutional encoder to extract features from the event representations, a bidirectional Gated Recurrent Unit (GRU), and a Linear Time-Varying State Space Module (LTV-SSM), to selectively capture contextual correlation from the forward and backward temporal relationship. Furthermore, the Bina-rep is utilized as a compact event representation, and the tailor-made data augmentation, called as Event-Cutout, is proposed to enhance the model's robustness by applying spatial random masking to the event image. The evaluation on the ThreeET-plus benchmark shows the superior performance of the MambaPupil, which secured the 1st place in CVPR'2024 AIS Event-based Eye Tracking challenge.
Authors: Junrui Zhang, Mozhgan Pourkeshavarz, Amir Rasouli
Abstract: As a safety critical task, autonomous driving requires accurate predictions of road users' future trajectories for safe motion planning, particularly under challenging conditions. Yet, many recent deep learning methods suffer from a degraded performance on the challenging scenarios, mainly because these scenarios appear less frequently in the training data. To address such a long-tail issue, existing methods force challenging scenarios closer together in the feature space during training to trigger information sharing among them for more robust learning. These methods, however, primarily rely on the motion patterns to characterize scenarios, omitting more informative contextual information, such as interactions and scene layout. We argue that exploiting such information not only improves prediction accuracy but also scene compliance of the generated trajectories. In this paper, we propose to incorporate richer training dynamics information into a prototypical contrastive learning framework. More specifically, we propose a two-stage process. First, we generate rich contextual features using a baseline encoder-decoder framework. These features are split into clusters based on the model's output errors, using the training dynamics information, and a prototype is computed within each cluster. Second, we retrain the model using the prototypes in a contrastive learning framework. We conduct empirical evaluations of our approach using two large-scale naturalistic datasets and show that our method achieves state-of-the-art performance by improving accuracy and scene compliance on the long-tail samples. Furthermore, we perform experiments on a subset of the clusters to highlight the additional benefit of our approach in reducing training bias.
Authors: Bingwen Zhu, Fanyi Wang, Tianyi Lu, Peng Liu, Jingwen Su, Jinxiu Liu, Yanhao Zhang, Zuxuan Wu, Yu-Gang Jiang, Guo-Jun Qi
Abstract: Image-to-video(I2V) generation aims to create a video sequence from a single image, which requires high temporal coherence and visual fidelity with the source image.However, existing approaches suffer from character appearance inconsistency and poor preservation of fine details. Moreover, they require a large amount of video data for training, which can be computationally demanding.To address these limitations,we propose PoseAnimate, a novel zero-shot I2V framework for character animation.PoseAnimate contains three key components: 1) Pose-Aware Control Module (PACM) incorporates diverse pose signals into conditional embeddings, to preserve character-independent content and maintain precise alignment of actions.2) Dual Consistency Attention Module (DCAM) enhances temporal consistency, and retains character identity and intricate background details.3) Mask-Guided Decoupling Module (MGDM) refines distinct feature perception, improving animation fidelity by decoupling the character and background.We also propose a Pose Alignment Transition Algorithm (PATA) to ensure smooth action transition.Extensive experiment results demonstrate that our approach outperforms the state-of-the-art training-based methods in terms of character consistency and detail fidelity. Moreover, it maintains a high level of temporal coherence throughout the generated animations.
Authors: Armando Zhu, Keqin Li, Tong Wu, Peng Zhao, Bo Hong
Abstract: With wearing masks becoming a new cultural norm, facial expression recognition (FER) while taking masks into account has become a significant challenge. In this paper, we propose a unified multi-branch vision transformer for facial expression recognition and mask wearing classification tasks. Our approach extracts shared features for both tasks using a dual-branch architecture that obtains multi-scale feature representations. Furthermore, we propose a cross-task fusion phase that processes tokens for each task with separate branches, while exchanging information using a cross attention module. Our proposed framework reduces the overall complexity compared with using separate networks for both tasks by the simple yet effective cross-task fusion phase. Extensive experiments demonstrate that our proposed model performs better than or on par with different state-of-the-art methods on both facial expression recognition and facial mask wearing classification task.
Authors: Ryoya Nara, Yu-Chieh Lin, Yuji Nozawa, Youyang Ng, Goh Itoh, Osamu Torii, Yusuke Matsui
Abstract: Many image retrieval studies use metric learning to train an image encoder. However, metric learning cannot handle differences in users' preferences, and requires data to train an image encoder. To overcome these limitations, we revisit relevance feedback, a classic technique for interactive retrieval systems, and propose an interactive CLIP-based image retrieval system with relevance feedback. Our retrieval system first executes the retrieval, collects each user's unique preferences through binary feedback, and returns images the user prefers. Even when users have various preferences, our retrieval system learns each user's preference through the feedback and adapts to the preference. Moreover, our retrieval system leverages CLIP's zero-shot transferability and achieves high accuracy without training. We empirically show that our retrieval system competes well with state-of-the-art metric learning in category-based image retrieval, despite not training image encoders specifically for each dataset. Furthermore, we set up two additional experimental settings where users have various preferences: one-label-based image retrieval and conditioned image retrieval. In both cases, our retrieval system effectively adapts to each user's preferences, resulting in improved accuracy compared to image retrieval without feedback. Overall, our work highlights the potential benefits of integrating CLIP with classic relevance feedback techniques to enhance image retrieval.
Authors: Zhe Chen, Weiyun Wang, Hao Tian, Shenglong Ye, Zhangwei Gao, Erfei Cui, Wenwen Tong, Kongzhi Hu, Jiapeng Luo, Zheng Ma, Ji Ma, Jiaqi Wang, Xiaoyi Dong, Hang Yan, Hewei Guo, Conghui He, Botian Shi, Zhenjiang Jin, Chao Xu, Bin Wang, Xingjian Wei, Wei Li, Wenjian Zhang, Bo Zhang, Pinlong Cai, Licheng Wen, Xiangchao Yan, Min Dou, Lewei Lu, Xizhou Zhu, Tong Lu, Dahua Lin, Yu Qiao, Jifeng Dai, Wenhai Wang
Abstract: In this report, we introduce InternVL 1.5, an open-source multimodal large language model (MLLM) to bridge the capability gap between open-source and proprietary commercial models in multimodal understanding. We introduce three simple improvements: (1) Strong Vision Encoder: we explored a continuous learning strategy for the large-scale vision foundation model -- InternViT-6B, boosting its visual understanding capabilities, and making it can be transferred and reused in different LLMs. (2) Dynamic High-Resolution: we divide images into tiles ranging from 1 to 40 of 448$\times$448 pixels according to the aspect ratio and resolution of the input images, which supports up to 4K resolution input. (3) High-Quality Bilingual Dataset: we carefully collected a high-quality bilingual dataset that covers common scenes, document images, and annotated them with English and Chinese question-answer pairs, significantly enhancing performance in OCR- and Chinese-related tasks. We evaluate InternVL 1.5 through a series of benchmarks and comparative studies. Compared to both open-source and proprietary models, InternVL 1.5 shows competitive performance, achieving state-of-the-art results in 8 of 18 benchmarks. Code has been released at https://github.com/OpenGVLab/InternVL.
Authors: Xiao Jiang, Fei Zhou
Abstract: Dynamic Vision Sensors (DVS) have recently generated great interest because of the advantages of wide dynamic range and low latency compared with conventional frame-based cameras. However, the complicated behaviors in dim light conditions are still not clear, restricting the applications of DVS. In this paper, we analyze the typical DVS circuit, and find that there exists discontinuity of event triggering time. In dim light conditions, the discontinuity becomes prominent. We point out that the discontinuity depends exclusively on the changing speed of light intensity. Experimental results on real event data validate the analysis and the existence of discontinuity that reveals the non-first-order behaviors of DVS in dim light conditions.
Authors: Pinxuan Dai, Jiamin Xu, Wenxiang Xie, Xinguo Liu, Huamin Wang, Weiwei Xu
Abstract: We propose a novel point-based representation, Gaussian surfels, to combine the advantages of the flexible optimization procedure in 3D Gaussian points and the surface alignment property of surfels. This is achieved by directly setting the z-scale of 3D Gaussian points to 0, effectively flattening the original 3D ellipsoid into a 2D ellipse. Such a design provides clear guidance to the optimizer. By treating the local z-axis as the normal direction, it greatly improves optimization stability and surface alignment. While the derivatives to the local z-axis computed from the covariance matrix are zero in this setting, we design a self-supervised normal-depth consistency loss to remedy this issue. Monocular normal priors and foreground masks are incorporated to enhance the quality of the reconstruction, mitigating issues related to highlights and background. We propose a volumetric cutting method to aggregate the information of Gaussian surfels so as to remove erroneous points in depth maps generated by alpha blending. Finally, we apply screened Poisson reconstruction method to the fused depth maps to extract the surface mesh. Experimental results show that our method demonstrates superior performance in surface reconstruction compared to state-of-the-art neural volume rendering and point-based rendering methods.
Authors: Zhuohao Li, Guoyang Xie, Guannan Jiang, Zhichao Lu
Abstract: Transformer recently emerged as the de facto model for computer vision tasks and has also been successfully applied to shadow removal. However, these existing methods heavily rely on intricate modifications to the attention mechanisms within the transformer blocks while using a generic patch embedding. As a result, it often leads to complex architectural designs requiring additional computation resources. In this work, we aim to explore the efficacy of incorporating shadow information within the early processing stage. Accordingly, we propose a transformer-based framework with a novel patch embedding that is tailored for shadow removal, dubbed ShadowMaskFormer. Specifically, we present a simple and effective mask-augmented patch embedding to integrate shadow information and promote the model's emphasis on acquiring knowledge for shadow regions. Extensive experiments conducted on the ISTD, ISTD+, and SRD benchmark datasets demonstrate the efficacy of our method against state-of-the-art approaches while using fewer model parameters.
Authors: Hyungyung Lee, Da Young Lee, Wonjae Kim, Jin-Hwa Kim, Tackeun Kim, Jihang Kim, Leonard Sunwoo, Edward Choi
Abstract: Synthetic medical data generation has opened up new possibilities in the healthcare domain, offering a powerful tool for simulating clinical scenarios, enhancing diagnostic and treatment quality, gaining granular medical knowledge, and accelerating the development of unbiased algorithms. In this context, we present a novel approach called ViewXGen, designed to overcome the limitations of existing methods that rely on general domain pipelines using only radiology reports to generate frontal-view chest X-rays. Our approach takes into consideration the diverse view positions found in the dataset, enabling the generation of chest X-rays with specific views, which marks a significant advancement in the field. To achieve this, we introduce a set of specially designed tokens for each view position, tailoring the generation process to the user's preferences. Furthermore, we leverage multi-view chest X-rays as input, incorporating valuable information from different views within the same study. This integration rectifies potential errors and contributes to faithfully capturing abnormal findings in chest X-ray generation. To validate the effectiveness of our approach, we conducted statistical analyses, evaluating its performance in a clinical efficacy metric on the MIMIC-CXR dataset. Also, human evaluation demonstrates the remarkable capabilities of ViewXGen, particularly in producing realistic view-specific X-rays that closely resemble the original images.
Authors: Soroush Abbasi Koohpayegani, KL Navaneet, Parsa Nooralinejad, Soheil Kolouri, Hamed Pirsiavash
Abstract: Fine-tuning Large Language Models (LLMs) and storing them for each downstream task or domain is impractical because of the massive model size (e.g., 350GB in GPT-3). Current literature, such as LoRA, showcases the potential of low-rank modifications to the original weights of an LLM, enabling efficient adaptation and storage for task-specific models. These methods can reduce the number of parameters needed to fine-tune an LLM by several orders of magnitude. Yet, these methods face two primary limitations: (1) the parameter count is lower-bounded by the rank one decomposition, and (2) the extent of reduction is heavily influenced by both the model architecture and the chosen rank. We introduce NOLA, which overcomes the rank one lower bound present in LoRA. It achieves this by re-parameterizing the low-rank matrices in LoRA using linear combinations of randomly generated matrices (basis) and optimizing the linear mixture coefficients only. This approach allows us to decouple the number of trainable parameters from both the choice of rank and the network architecture. We present adaptation results using GPT-2, LLaMA-2, and ViT in natural language and computer vision tasks. NOLA performs as well as LoRA models with much fewer number of parameters compared to LoRA with rank one, the best compression LoRA can archive. Particularly, on LLaMA-2 70B, our method is almost 20 times more compact than the most compressed LoRA without degradation in accuracy. Our code is available here: https://github.com/UCDvision/NOLA
Authors: Tanner J. Zachem (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA, Department of Mechanical Engineering and Materials Science, Duke University, Durham, NC, USA), Sully F. Chen (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA), Vishal Venkatraman (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA), David AW Sykes (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA), Ravi Prakash (Department of Mechanical Engineering and Materials Science, Duke University, Durham, NC, USA), Koumani W. Ntowe (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA), Mikhail A. Bethell (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA), Samantha Spellicy (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA), Alexander D Suarez (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA), Weston Ross (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA), Patrick J. Codd (Department of Neurosurgery, Duke University School of Medicine, Durham, NC, USA, Department of Mechanical Engineering and Materials Science, Duke University, Durham, NC, USA)
Abstract: Objectives Computer vision (CV) is a field of artificial intelligence that enables machines to interpret and understand images and videos. CV has the potential to be of assistance in the operating room (OR) to track surgical instruments. We built a CV algorithm for identifying surgical instruments in the neurosurgical operating room as a potential solution for surgical instrument tracking and management to decrease surgical waste and opening of unnecessary tools. Methods We collected 1660 images of 27 commonly used neurosurgical instruments. Images were labeled using the VGG Image Annotator and split into 80% training and 20% testing sets in order to train a U-Net Convolutional Neural Network using 5-fold cross validation. Results Our U-Net achieved a tool identification accuracy of 80-100% when distinguishing 25 classes of instruments, with 19/25 classes having accuracy over 90%. The model performance was not adequate for sub classifying Adson, Gerald, and Debakey forceps, which had accuracies of 60-80%. Conclusions We demonstrated the viability of using machine learning to accurately identify surgical instruments. Instrument identification could help optimize surgical tray packing, decrease tool usage and waste, decrease incidence of instrument misplacement events, and assist in timing of routine instrument maintenance. More training data will be needed to increase accuracy across all surgical instruments that would appear in a neurosurgical operating room. Such technology has the potential to be used as a method to be used for proving what tools are truly needed in each type of operation allowing surgeons across the world to do more with less.
Authors: Hadis Keshavarz, Hossein Sadr
Abstract: In the era of digital medicine, medical imaging serves as a widespread technique for early disease detection, with a substantial volume of images being generated and stored daily in electronic patient records. X-ray angiography imaging is a standard and one of the most common methods for rapidly diagnosing coronary artery diseases. The notable achievements of recent deep learning algorithms align with the increased use of electronic health records and diagnostic imaging. Deep neural networks, leveraging abundant data, advanced algorithms, and powerful computational capabilities, prove highly effective in the analysis and interpretation of images. In this context, Object detection methods have become a promising approach, particularly through convolutional neural networks (CNN), streamlining medical image analysis by eliminating manual feature extraction. This allows for direct feature extraction from images, ensuring high accuracy in results. Therefore, in our paper, we utilized the object detection method on X-ray angiography images to precisely identify the location of coronary artery stenosis. As a result, this model enables automatic and real-time detection of stenosis locations, assisting in the crucial and sensitive decision-making process for healthcare professionals.
Authors: Hyenkyun Woo
Abstract: This article presents a new polynomial parameterized sigmoid called SIGTRON, which is an extended asymmetric sigmoid with Perceptron, and its companion convex model called SIGTRON-imbalanced classification (SIC) model that employs a virtual SIGTRON-induced convex loss function. In contrast to the conventional $\pi$-weighted cost-sensitive learning model, the SIC model does not have an external $\pi$-weight on the loss function but has internal parameters in the virtual SIGTRON-induced loss function. As a consequence, when the given training dataset is close to the well-balanced condition considering the (scale-)class-imbalance ratio, we show that the proposed SIC model is more adaptive to variations of the dataset, such as the inconsistency of the (scale-)class-imbalance ratio between the training and test datasets. This adaptation is justified by a skewed hyperplane equation, created via linearization of the gradient satisfying $\epsilon$-optimal condition. Additionally, we present a quasi-Newton optimization(L-BFGS) framework for the virtual convex loss by developing an interval-based bisection line search. Empirically, we have observed that the proposed approach outperforms (or is comparable to) $\pi$-weighted convex focal loss and balanced classifier LIBLINEAR(logistic regression, SVM, and L2SVM) in terms of test classification accuracy with $51$ two-class and $67$ multi-class datasets. In binary classification problems, where the scale-class-imbalance ratio of the training dataset is not significant but the inconsistency exists, a group of SIC models with the best test accuracy for each dataset (TOP$1$) outperforms LIBSVM(C-SVC with RBF kernel), a well-known kernel-based classifier.
Authors: Hubert Padusinski, Christian Steinhauser, Thilo Braun, Lennart Ries, Eric Sax
Abstract: Machine Vision (MV) is essential for solving driving automation. This paper examines potential shortcomings in current MV testing strategies for highly automated driving (HAD) systems. We argue for a more comprehensive understanding of the performance factors that must be considered during the MV evaluation process, noting that neglecting these factors can lead to significant risks. This is not only relevant to MV component testing, but also to integration testing. To illustrate this point, we draw an analogy to a ship navigating towards an iceberg to show potential hidden challenges in current MV testing strategies. The main contribution is a novel framework for black-box testing which observes environmental relations. This means it is designed to enhance MV assessments by considering the attributes and surroundings of relevant individual objects. The framework provides the identification of seven general concerns about the object recognition of MV, which are not addressed adequately in established test processes. To detect these deficits based on their performance factors, we propose the use of a taxonomy called "granularity orders" along with a graphical representation. This allows an identification of MV uncertainties across a range of driving scenarios. This approach aims to advance the precision, efficiency, and completeness of testing procedures for MV.
Authors: Qiyao Liang, Ziming Liu, Ila Fiete
Abstract: Diffusion models are capable of impressive feats of image generation with uncommon juxtapositions such as astronauts riding horses on the moon with properly placed shadows. These outputs indicate the ability to perform compositional generalization, but how do the models do so? We perform controlled experiments on conditional DDPMs learning to generate 2D spherical Gaussian bumps centered at specified $x$- and $y$-positions. Our results show that the emergence of semantically meaningful latent representations is key to achieving high performance. En route to successful performance over learning, the model traverses three distinct phases of latent representations: (phase A) no latent structure, (phase B) a 2D manifold of disordered states, and (phase C) a 2D ordered manifold. Corresponding to each of these phases, we identify qualitatively different generation behaviors: 1) multiple bumps are generated, 2) one bump is generated but at inaccurate $x$ and $y$ locations, 3) a bump is generated at the correct $x$ and y location. Furthermore, we show that even under imbalanced datasets where features ($x$- versus $y$-positions) are represented with skewed frequencies, the learning process for $x$ and $y$ is coupled rather than factorized, demonstrating that simple vanilla-flavored diffusion models cannot learn efficient representations in which localization in $x$ and $y$ are factorized into separate 1D tasks. These findings suggest the need for future work to find inductive biases that will push generative models to discover and exploit factorizable independent structures in their inputs, which will be required to vault these models into more data-efficient regimes.
Authors: Yidong Zhao, Yi Zhang, Qian Tao
Abstract: Deep learning-based methods have achieved prestigious performance for magnetic resonance imaging (MRI) reconstruction, enabling fast imaging for many clinical applications. Previous methods employ convolutional networks to learn the image prior as the regularization term. In quantitative MRI, the physical model of nuclear magnetic resonance relaxometry is known, providing additional prior knowledge for image reconstruction. However, traditional reconstruction networks are limited to learning the spatial domain prior knowledge, ignoring the relaxometry prior. Therefore, we propose a relaxometry-guided quantitative MRI reconstruction framework to learn the spatial prior from data and the relaxometry prior from MRI physics. Additionally, we also evaluated the performance of two popular reconstruction backbones, namely, recurrent variational networks (RVN) and variational networks (VN) with U- Net. Experiments demonstrate that the proposed method achieves highly promising results in quantitative MRI reconstruction.
Authors: Minhyuk Seo, Diganta Misra, Seongwon Cho, Minjae Lee, Jonghyun Choi
Abstract: In real-world scenarios, extensive manual annotation for continual learning is impractical due to prohibitive costs. Although prior arts, influenced by large-scale webly supervised training, suggest leveraging web-scraped data in continual learning, this poses challenges such as data imbalance, usage restrictions, and privacy concerns. Addressing the risks of continual webly supervised training, we present an online continual learning framework - Generative Name only Continual Learning (G-NoCL). The proposed G-NoCL uses a set of generators G along with the learner. When encountering new concepts (i.e., classes), G-NoCL employs the novel sample complexity-guided data ensembling technique DIverSity and COmplexity enhancing ensemBlER (DISCOBER) to optimally sample training data from generated data. Through extensive experimentation, we demonstrate superior performance of DISCOBER in G-NoCL online CL benchmarks, covering both In-Distribution (ID) and Out-of-Distribution (OOD) generalization evaluations, compared to naive generator-ensembling, web-supervised, and manually annotated data.
Authors: Ye Wang, Yaxiong Wang, Yujiao Wu, Bingchen Zhao, Xueming Qian
Abstract: Generalized Class Discovery (GCD) aims to dynamically assign labels to unlabelled data partially based on knowledge learned from labelled data, where the unlabelled data may come from known or novel classes. The prevailing approach generally involves clustering across all data and learning conceptions by prototypical contrastive learning. However, existing methods largely hinge on the performance of clustering algorithms and are thus subject to their inherent limitations. Firstly, the estimated cluster number is often smaller than the ground truth, making the existing methods suffer from the lack of prototypes for comprehensive conception learning. To address this issue, we propose an adaptive probing mechanism that introduces learnable potential prototypes to expand cluster prototypes (centers). As there is no ground truth for the potential prototype, we develop a self-supervised prototype learning framework to optimize the potential prototype in an end-to-end fashion. Secondly, clustering is computationally intensive, and the conventional strategy of clustering both labelled and unlabelled instances exacerbates this issue. To counteract this inefficiency, we opt to cluster only the unlabelled instances and subsequently expand the cluster prototypes with our introduced potential prototypes to fast explore novel classes. Despite the simplicity of our proposed method, extensive empirical analysis on a wide range of datasets confirms that our method consistently delivers state-of-the-art results. Specifically, our method surpasses the nearest competitor by a significant margin of 9.7% within the Stanford Cars dataset and 12x clustering efficiency within the Herbarium 19 dataset. We will make the code and checkpoints publicly available at https://github.com/xjtuYW/PNP.git.