Authors: Yang Zhou, Tao Huang, Ravi Ramamoorthi, Pradeep Sen, Ling-Qi Yan
Abstract: Creating an appearance-preserving level-of-detail (LoD) representation for arbitrary 3D scenes is a challenging problem. The appearance of a scene is an intricate combination of both geometry and material models, and is further complicated by correlation due to the spatial configuration of scene elements. We present a novel volumetric representation for the aggregated appearance of complex scenes and an efficient pipeline for LoD generation and rendering. The core of our representation is the Aggregated Bidirectional Scattering Distribution Function (ABSDF) that summarizes the far-field appearance of all surfaces inside a voxel. We propose a closed-form factorization of the ABSDF that accounts for spatially varying and orientation-varying material parameters. We tackle the challenge of capturing the correlation existing locally within a voxel and globally across different parts of the scene. Our method faithfully reproduces appearance and achieves higher quality than existing scene filtering methods while being inherently efficient to render. The memory footprint and rendering cost of our representation are independent of the original scene complexity.
Authors: Farshid Ghezelbash, Amir Hossein Eskandari, Amir J Bidhendi
Abstract: This study introduces a dataset consisting of approximately 9,000 images of mechanical mechanisms and their corresponding descriptions, aimed at supporting research in mechanism design. The dataset consists of a diverse collection of 2D and 3D sketches, meticulously curated to ensure relevance and quality. We demonstrate the application of this dataset by fine-tuning two models: 1) Stable Diffusion (for generating new mechanical designs), and 2) BLIP-2 (for captioning these designs). While the results from Stable Diffusion show promise, particularly in generating coherent 3D sketches, the model struggles with 2D sketches and occasionally produces nonsensical outputs. These limitations underscore the need for further development, particularly in expanding the dataset and refining model architectures. Nonetheless, this work serves as a step towards leveraging generative AI in mechanical design, highlighting both the potential and current limitations of these approaches.
Authors: Nilay Kant, Ashrut Aryal, Rajiv Ranganathan, Ranjan Mukherjee, Charles Owen
Abstract: This paper explores a novel approach to model strategies for flattening wrinkled cloth learning from humans. A human participant study was conducted where the participants were presented with various wrinkle types and tasked with flattening the cloth using the fewest actions possible. A camera and Aruco marker were used to capture images of the cloth and finger movements, respectively. The human strategies for flattening the cloth were modeled using a supervised regression neural network, where the cloth images served as input and the human actions as output. Before training the neural network, a series of image processing techniques were applied, followed by Principal Component Analysis (PCA) to extract relevant features from each image and reduce the input dimensionality. This reduction decreased the model's complexity and computational cost. The actions predicted by the neural network closely matched the actual human actions on an independent data set, demonstrating the effectiveness of neural networks in modeling human actions for flattening wrinkled cloth.
Authors: Martin Obschonka, Christian Fisch, Tharindu Fernando, Clinton Fookes
Abstract: Occupational outcomes like entrepreneurship are generally considered personal information that individuals should have the autonomy to disclose. With the advancing capability of artificial intelligence (AI) to infer private details from widely available human-centric data, such as social media, it is crucial to investigate whether AI can accurately extract private occupational information from such data. In this study, we demonstrate that deep neural networks can classify individuals as entrepreneurs based on a single facial image with high accuracy in data sourced from Crunchbase, a premier source for entrepreneurship data. Utilizing a dataset comprising facial images of 40,728 individuals, including both entrepreneurs and non-entrepreneurs, we trained a Convolutional Neural Network (CNN) and evaluated its classification performance. While human experts (n=650) and trained participants (n=133) were unable to classify entrepreneurs with accuracy above chance levels (>50%), the AI model achieved a classification accuracy of 79.51%. Several robustness tests show that this high level of accuracy is maintained under various conditions.
Authors: Reza Kakavand, Reza Ahmadi, Atousa Parsaei, W. Brent Edwards, Amin Komeili
Abstract: Markerless motion capture offers several benefits over traditional marker-based systems by eliminating the need for physical markers, which are prone to misplacement and artifacts. Utilizing computer vision and deep learning algorithms, markerless systems can directly detect human body landmarks, reducing manual processing and errors associated with marker placement. These systems are adaptable, able to track user-defined features, and practical for real-world applications using consumer-grade devices such as smartphone cameras. This study compares the performance of OpenCap, a markerless motion capture system, with traditional marker-based systems in assessing cycling biomechanics. Ten healthy adults participated in experiments to capture sagittal hip, knee, and ankle kinematics and dynamics using both methods. OpenCap used videos from smartphones and integrated computer vision and musculoskeletal simulations to estimate 3D kinematics. Results showed high agreement between the two systems, with no significant differences in kinematic and kinetic measurements for the hip, knee, and ankle. The correlation coefficients exceeded 0.98, indicating very strong consistency. Errors were minimal, with kinematic errors under 4 degrees and kinetic errors below 5 Nm. This study concludes that OpenCap is a viable alternative to marker-based motion capture, offering comparable precision without extensive setup for hip (flexion/extension), knee (flexion/extension), and ankle (dorsiflexion/plantarflexion) joints. Future work should aim to enhance the accuracy of ankle joint measurements and extend analyses to 3D kinematics and kinetics for comprehensive biomechanical assessments.
Authors: Sakhinana Sagar Srinivas, Rajat Kumar Sarkar, Venkataramana Runkana
Abstract: Characterization of materials via electron micrographs is an important and challenging task in several materials processing industries. Classification of electron micrographs is complex due to the high intra-class dissimilarity, high inter-class similarity, and multi-spatial scales of patterns. However, existing methods are ineffective in learning complex image patterns. We propose an effective end-to-end electron micrograph representation learning-based framework for nanomaterial identification to overcome the challenges. We demonstrate that our framework outperforms the popular baselines on the open-source datasets in nanomaterials-based identification tasks. The ablation studies are reported in great detail to support the efficacy of our approach.
Authors: Kiran Purohit, Anurag Reddy Parvathgari, Sourangshu Bhattacharya
Abstract: Deep convolutional neural networks (CNNs) have achieved impressive performance in many computer vision tasks. However, their large model sizes require heavy computational resources, making pruning redundant filters from existing pre-trained CNNs an essential task in developing efficient models for resource-constrained devices. Whole-network filter pruning algorithms prune varying fractions of filters from each layer, hence providing greater flexibility. Current whole-network pruning methods are either computationally expensive due to the need to calculate the loss for each pruned filter using a training dataset, or use various heuristic / learned criteria for determining the pruning fractions for each layer. This paper proposes a two-level hierarchical approach for whole-network filter pruning which is efficient and uses the classification loss as the final criterion. The lower-level algorithm (called filter-pruning) uses a sparse-approximation formulation based on linear approximation of filter weights. We explore two algorithms: orthogonal matching pursuit-based greedy selection and a greedy backward pruning approach. The backward pruning algorithm uses a novel closed-form error criterion for efficiently selecting the optimal filter at each stage, thus making the whole algorithm much faster. The higher-level algorithm (called layer-selection) greedily selects the best-pruned layer (pruning using the filter-selection algorithm) using a global pruning criterion. We propose algorithms for two different global-pruning criteria: (1) layer-wise relative error (HBGS), and (2) final classification error (HBGTS). Our suite of algorithms outperforms state-of-the-art pruning methods on ResNet18, ResNet32, ResNet56, VGG16, and ResNext101. Our method reduces the RAM requirement for ResNext101 from 7.6 GB to 1.5 GB and achieves a 94% reduction in FLOPS without losing accuracy on CIFAR-10.
Authors: Chengjie Lu, Jiahui Wu, Shaukat Ali, Mikkel Labori Olsen
Abstract: Refurbishing laptops extends their lives while contributing to reducing electronic waste, which promotes building a sustainable future. To this end, the Danish Technological Institute (DTI) focuses on the research and development of several applications, including laptop refurbishing. This has several steps, including cleaning, which involves identifying and removing stickers from laptop surfaces. DTI trained six sticker detection models (SDMs) based on open-source object detection models to identify such stickers precisely so these stickers can be removed automatically. However, given the diversity in types of stickers (e.g., shapes, colors, locations), identification of the stickers is highly uncertain, thereby requiring explicit quantification of uncertainty associated with the identified stickers. Such uncertainty quantification can help reduce risks in removing stickers, which, for example, could otherwise result in damaging laptop surfaces. For uncertainty quantification, we adopted the Monte Carlo Dropout method to evaluate the six SDMs from DTI using three datasets: the original image dataset from DTI and two datasets generated with vision language models, i.e., DALL-E-3 and Stable Diffusion-3. In addition, we presented novel robustness metrics concerning detection accuracy and uncertainty to assess the robustness of the SDMs based on adversarial datasets generated from the three datasets using a dense adversary method. Our evaluation results show that different SDMs perform differently regarding different metrics. Based on the results, we provide SDM selection guidelines and lessons learned from various perspectives.
Authors: Fereshteh Shakeri, Yunshi Huang, Julio Silva-Rodr\'iguez, Houda Bahig, An Tang, Jose Dolz, Ismail Ben Ayed
Abstract: Integrating image and text data through multi-modal learning has emerged as a new approach in medical imaging research, following its successful deployment in computer vision. While considerable efforts have been dedicated to establishing medical foundation models and their zero-shot transfer to downstream tasks, the popular few-shot setting remains relatively unexplored. Following on from the currently strong emergence of this setting in computer vision, we introduce the first structured benchmark for adapting medical vision-language models (VLMs) in a strict few-shot regime and investigate various adaptation strategies commonly used in the context of natural images. Furthermore, we evaluate a simple generalization of the linear-probe adaptation baseline, which seeks an optimal blending of the visual prototypes and text embeddings via learnable class-wise multipliers. Surprisingly, such a text-informed linear probe yields competitive performances in comparison to convoluted prompt-learning and adapter-based strategies, while running considerably faster and accommodating the black-box setting. Our extensive experiments span three different medical modalities and specialized foundation models, nine downstream tasks, and several state-of-the-art few-shot adaptation methods. We made our benchmark and code publicly available to trigger further developments in this emergent subject: \url{https://github.com/FereshteShakeri/few-shot-MedVLMs}.
Authors: Zhuang Jia, Wenkai Lu, Meng Zhang, Yongkang Miao
Abstract: Ground-roll wave is a common coherent noise in land field seismic data. This Rayleigh-type surface wave usually has low frequency, low apparent velocity, and high amplitude, therefore obscures the reflection events of seismic shot gathers. Commonly used techniques focus on the differences of ground-roll and reflection in transformed domain such as $f-k$ domain, wavelet domain, or curvelet domain. These approaches use a series of fixed atoms or bases to transform the data in time-space domain into transformed domain to separate different waveforms, thus tend to suffer from the complexity for a delicate design of the parameters of the transform domain filter. To deal with these problems, a novel way is proposed to separate ground-roll from reflections using convolutional neural network (CNN) model based method to learn to extract the features of ground-roll and reflections automatically based on training data. In the proposed method, low-pass filtered seismic data which is contaminated by ground-roll wave is used as input of CNN, and then outputs both ground-roll component and low-frequency part of reflection component simultaneously. Discriminative loss is applied together with similarity loss in the training process to enhance the similarity to their train labels as well as the difference between the two outputs. Experiments are conducted on both synthetic and real data, showing that CNN based method can separate ground roll from reflections effectively, and has generalization ability to a certain extent.
Authors: Federico Cunico, Marco Cristani
Abstract: In recent years, the development of deep learning approaches for the task of person re-identification led to impressive results. However, this comes with a limitation for industrial and practical real-world applications. Firstly, most of the existing works operate on closed-world scenarios, in which the people to re-identify (probes) are compared to a closed-set (gallery). Real-world scenarios often are open-set problems in which the gallery is not known a priori, but the number of open-set approaches in the literature is significantly lower. Secondly, challenges such as multi-camera setups, occlusions, real-time requirements, etc., further constrain the applicability of off-the-shelf methods. This work presents MICRO-TRACK, a Modular Industrial multi-Camera Re_identification and Open-set Tracking system that is real-time, scalable, and easy to integrate into existing industrial surveillance scenarios. Furthermore, we release a novel Re-ID and tracking dataset acquired in an industrial manufacturing facility, dubbed Facility-ReID, consisting of 18-minute videos captured by 8 surveillance cameras.
Authors: Arnold Schwarz, Levente Hernadi, Felix Bie{\ss}mann, Kristian Hildebrand
Abstract: In this study we provide empirical evidence demonstrating that the quality of training data impacts model performance in Human Pose Estimation (HPE). Inaccurate labels in widely used data sets, ranging from minor errors to severe mislabeling, can negatively influence learning and distort performance metrics. We perform an in-depth analysis of popular HPE data sets to show the extent and nature of label inaccuracies. Our findings suggest that accounting for the impact of faulty labels will facilitate the development of more robust and accurate HPE models for a variety of real-world applications. We show improved performance with cleansed data.
Authors: Mallika Garg, Debashis Ghosh, Pyari Mohan Pradhan
Abstract: In this paper, we introduce a novel Multiscale Video Transformer Network (MVTN) for dynamic hand gesture recognition, since multiscale features can extract features with variable size, pose, and shape of hand which is a challenge in hand gesture recognition. The proposed model incorporates a multiscale feature hierarchy to capture diverse levels of detail and context within hand gestures which enhances the model's ability. This multiscale hierarchy is obtained by extracting different dimensions of attention in different transformer stages with initial stages to model high-resolution features and later stages to model low-resolution features. Our approach also leverages multimodal data, utilizing depth maps, infrared data, and surface normals along with RGB images from NVGesture and Briareo datasets. Experiments show that the proposed MVTN achieves state-of-the-art results with less computational complexity and parameters. The source code is available at https://github.com/mallikagarg/MVTN.
Authors: Thanh-Dung Le, Vu Nguyen Ha, Ti Ti Nguyen, Geoffrey Eappen, Prabhu Thiruvasagam, Luis M. Garces-Socarras, Hong-fu Chou, Jorge L. Gonzalez-Rios, Juan Carlos Merlano-Duncan, Symeon Chatzinotas
Abstract: Remote sensing image classification is a critical component of Earth observation (EO) systems, traditionally dominated by convolutional neural networks (CNNs) and other deep learning techniques. However, the advent of Transformer-based architectures and large-scale pre-trained models has significantly shifted, offering enhanced performance and efficiency. This study focuses on identifying the most effective pre-trained model for land use classification in onboard satellite processing, emphasizing achieving high accuracy, computational efficiency, and robustness against noisy data conditions commonly encountered during satellite-based inference. Through extensive experimentation, we compared traditional CNN-based models, ResNet-based models, and various pre-trained vision Transformer models. Our findings demonstrate that pre-trained Transformer models, particularly MobileViTV2 and EfficientViT-M2, outperform models trained from scratch in accuracy and efficiency. These models achieve high performance with reduced computational requirements and exhibit greater resilience during inference under noisy conditions. While MobileViTV2 excelled on clean validation data, EfficientViT-M2 proved more robust when handling noise, making it the most suitable model for onboard satellite Earth observation tasks. In conclusion, EfficientViT-M2 is the optimal choice for reliable and efficient remote sensing image classification in satellite operations, achieving 98.76\% accuracy, precision, and recall. Specifically, EfficientViT-M2 delivered the highest performance across all metrics, excelled in training efficiency (1,000s) and inference time (10s), and demonstrated greater robustness (overall robustness score at 0.79).
Authors: \`Eric \'Sanchez, Adri\`a Molina, Oriol Ramos Terrades
Abstract: The use of image analysis in automated photography management is an increasing trend in heritage institutions. Such tools alleviate the human cost associated with the manual and expensive annotation of new data sources while facilitating fast access to the citizenship through online indexes and search engines. However, available tagging and description tools are usually designed around modern photographs in English, neglecting historical corpora in minoritized languages, each of which exhibits intrinsic particularities. The primary objective of this research is to study the quantitative contribution of generative systems in the description of historical sources. This is done by contextualizing the task of captioning historical photographs from the Catalan archives as a case study. Our findings provide practitioners with tools and directions on transfer learning for captioning models based on visual adaptation and linguistic proximity.
Authors: Jenil Kanani
Abstract: The exponential growth in waste production due to rapid economic and industrial development necessitates efficient waste management strategies to mitigate environmental pollution and resource depletion. Leveraging advancements in computer vision, this study proposes a novel approach inspired by pixel distribution learning techniques to enhance automated garbage classification. The method aims to address limitations of conventional convolutional neural network (CNN)-based approaches, including computational complexity and vulnerability to image variations. We will conduct experiments using the Kaggle Garbage Classification dataset, comparing our approach with existing models to demonstrate the strength and efficiency of pixel distribution learning in automated garbage classification technologies.
Authors: Zhe Li, Weitong Zhang, Sarah Cechnicka, Bernhard Kainz
Abstract: While deep learning techniques have proven successful in image-related tasks, the exponentially increased data storage and computation costs become a significant challenge. Dataset distillation addresses these challenges by synthesizing only a few images for each class that encapsulate all essential information. Most current methods focus on matching. The problems lie in the synthetic images not being human-readable and the dataset performance being insufficient for downstream learning tasks. Moreover, the distillation time can quickly get out of bounds when the number of synthetic images per class increases even slightly. To address this, we train a class conditional latent diffusion model capable of generating realistic synthetic images with labels. The sampling time can be reduced to several tens of images per seconds. We demonstrate that models can be effectively trained using only a small set of synthetic images and evaluated on a large real test set. Our approach achieved rank \(1\) in The First Dataset Distillation Challenge at ECCV 2024 on the CIFAR100 and TinyImageNet datasets.
Authors: Isaac Ray, Alexei Skurikhin
Abstract: This paper proposes a method for unsupervised whole-image clustering of a target dataset of remote sensing scenes with no labels. The method consists of three main steps: (1) finetuning a pretrained deep neural network (DINOv2) on a labelled source remote sensing imagery dataset and using it to extract a feature vector from each image in the target dataset, (2) reducing the dimension of these deep features via manifold projection into a low-dimensional Euclidean space, and (3) clustering the embedded features using a Bayesian nonparametric technique to infer the number and membership of clusters simultaneously. The method takes advantage of heterogeneous transfer learning to cluster unseen data with different feature and label distributions. We demonstrate the performance of this approach outperforming state-of-the-art zero-shot classification methods on several remote sensing scene classification datasets.
Authors: Shashank Tripathi, Omid Taheri, Christoph Lassner, Michael J. Black, Daniel Holden, Carsten Stoll
Abstract: Generating realistic human motion is essential for many computer vision and graphics applications. The wide variety of human body shapes and sizes greatly impacts how people move. However, most existing motion models ignore these differences, relying on a standardized, average body. This leads to uniform motion across different body types, where movements don't match their physical characteristics, limiting diversity. To solve this, we introduce a new approach to develop a generative motion model based on body shape. We show that it's possible to train this model using unpaired data by applying cycle consistency, intuitive physics, and stability constraints, which capture the relationship between identity and movement. The resulting model generates diverse, physically plausible, and dynamically stable human motions that are both quantitatively and qualitatively more realistic than current state-of-the-art methods. More details are available on our project page https://CarstenEpic.github.io/humos/.
Authors: Konstantinos Fotopoulos, Petros Maragos, Panagiotis Misiakos
Abstract: We present TropNNC, a structured pruning framework for compressing neural networks with linear and convolutional layers and ReLU activations. Our approximation is based on a geometrical approach to machine/deep learning, using tropical geometry and extending the work of Misiakos et al. (2022). We use the Hausdorff distance of zonotopes in its standard continuous form to achieve a tighter approximation bound for tropical polynomials compared to Misiakos et al. (2022). This enhancement allows for superior functional approximations of neural networks, leading to a more effective compression algorithm. Our method is significantly easier to implement compared to other frameworks, and does not depend on the availability of training data samples. We validate our framework through extensive empirical evaluations on the MNIST, CIFAR, and ImageNet datasets. Our results demonstrate that TropNNC achieves performance on par with the state-of-the-art method ThiNet, even surpassing it in compressing linear layers, and to the best of our knowledge, it is the first method that achieves this using tropical geometry.
Authors: Kai Shu, Yuzhuo Jia, Ziyang Zhang, Jiechao Gao
Abstract: Automatic Medical Imaging Narrative generation aims to alleviate the workload of radiologists by producing accurate clinical descriptions directly from radiological images. However, the subtle visual nuances and domain-specific terminology in medical images pose significant challenges compared to generic image captioning tasks. Existing approaches often neglect the vital distinction between normal and abnormal findings, leading to suboptimal performance. In this work, we propose FODA-PG, a novel Fine-grained Organ-Disease Adaptive Partitioning Graph framework that addresses these limitations through domain-adaptive learning. FODA-PG constructs a granular graphical representation of radiological findings by separating disease-related attributes into distinct "disease-specific" and "disease-free" categories based on their clinical significance and location. This adaptive partitioning enables our model to capture the nuanced differences between normal and pathological states, mitigating the impact of data biases. By integrating this fine-grained semantic knowledge into a powerful transformer-based architecture and providing rigorous mathematical justifications for its effectiveness, FODA-PG generates precise and clinically coherent reports with enhanced generalization capabilities. Extensive experiments on the IU-Xray and MIMIC-CXR benchmarks demonstrate the superiority of our approach over state-of-the-art methods, highlighting the importance of domain adaptation in medical report generation.
Authors: Tahsina Hashem, Weiqing Wang, Derry Tanti Wijaya, Mohammed Eunus Ali, Yuan-Fang Li
Abstract: While large multimodal models (LMMs) have obtained strong performance on many multimodal tasks, they may still hallucinate while generating text. Their performance on detecting salient features from visual data is also unclear. In this paper, we develop a framework to generate faithful and salient text from mixed-modal data, which includes images and structured data ( represented in knowledge graphs or tables). Specifically, we train a small vision critic model to identify hallucinated and non-salient features from the image modality. The critic model also generates a list of salient image features. This information is used in the post editing step to improve the generation quality. Experiments on two datasets show that our framework improves LMMs' generation quality on both faithfulness and saliency, outperforming recent techniques aimed at reducing hallucination.
Authors: Dongping Zhang, Zheng Li, Fangao Zeng, Yutong Wei
Abstract: Tooth segmentation is a critical technology in the field of medical image segmentation, with applications ranging from orthodontic treatment to human body identification and dental pathology assessment. Despite the development of numerous tooth image segmentation models by researchers, a common shortcoming is the failure to account for the challenges of blurred tooth boundaries. Dental diagnostics require precise delineation of tooth boundaries. This paper introduces an innovative tooth segmentation network that integrates boundary information to address the issue of indistinct boundaries between teeth and adjacent tissues. This network's core is its boundary feature extraction module, which is designed to extract detailed boundary information from high-level features. Concurrently, the feature cross-fusion module merges detailed boundary and global semantic information in a synergistic way, allowing for stepwise layer transfer of feature information. This method results in precise tooth segmentation. In the most recent STS Data Challenge, our methodology was rigorously tested and received a commendable overall score of 0.91. When compared to other existing approaches, this score demonstrates our method's significant superiority in segmenting tooth boundaries.
Authors: Jianbiao Mei, Yukai Ma, Xuemeng Yang, Licheng Wen, Tiantian Wei, Min Dou, Botian Shi, Yong Liu
Abstract: Recent advances in diffusion models have significantly enhanced the cotrollable generation of streetscapes for and facilitated downstream perception and planning tasks. However, challenges such as maintaining temporal coherence, generating long videos, and accurately modeling driving scenes persist. Accordingly, we propose DreamForge, an advanced diffusion-based autoregressive video generation model designed for the long-term generation of 3D-controllable and extensible video. In terms of controllability, our DreamForge supports flexible conditions such as text descriptions, camera poses, 3D bounding boxes, and road layouts, while also providing perspective guidance to produce driving scenes that are both geometrically and contextually accurate. For consistency, we ensure inter-view consistency through cross-view attention and temporal coherence via an autoregressive architecture enhanced with motion cues. Codes will be available at https://github.com/PJLab-ADG/DriveArena.
Authors: Gang Dai, Yifan Zhang, Quhui Ke, Qiangya Guo, Shuangping Huang
Abstract: Existing handwritten text generation methods often require more than ten handwriting samples as style references. However, in practical applications, users tend to prefer a handwriting generation model that operates with just a single reference sample for its convenience and efficiency. This approach, known as "one-shot generation", significantly simplifies the process but poses a significant challenge due to the difficulty of accurately capturing a writer's style from a single sample, especially when extracting fine details from the characters' edges amidst sparse foreground and undesired background noise. To address this problem, we propose a One-shot Diffusion Mimicker (One-DM) to generate handwritten text that can mimic any calligraphic style with only one reference sample. Inspired by the fact that high-frequency information of the individual sample often contains distinct style patterns (e.g., character slant and letter joining), we develop a novel style-enhanced module to improve the style extraction by incorporating high-frequency components from a single sample. We then fuse the style features with the text content as a merged condition for guiding the diffusion model to produce high-quality handwritten text images. Extensive experiments demonstrate that our method can successfully generate handwriting scripts with just one sample reference in multiple languages, even outperforming previous methods using over ten samples. Our source code is available at https://github.com/dailenson/One-DM.
Authors: Jing Wang, Ao Ma, Jiasong Feng, Dawei Leng, Yuhui Yin, Xiaodan Liang
Abstract: The global self-attention mechanism in diffusion transformers involves redundant computation due to the sparse and redundant nature of visual information, and the attention map of tokens within a spatial window shows significant similarity. To address this redundancy, we propose the Proxy Token Diffusion Transformer (PT-DiT), which employs sparse representative token attention (where the number of representative tokens is much smaller than the total number of tokens) to model global visual information efficiently. Specifically, in each transformer block, we randomly sample one token from each spatial-temporal window to serve as a proxy token for that region. The global semantics are captured through the self-attention of these proxy tokens and then injected into all latent tokens via cross-attention. Simultaneously, we introduce window and shift window attention to address the limitations in detail modeling caused by the sparse attention mechanism. Building on the well-designed PT-DiT, we further develop the Qihoo-T2X family, which includes a variety of models for T2I, T2V, and T2MV tasks. Experimental results show that PT-DiT achieves competitive performance while reducing the computational complexity in both image and video generation tasks (e.g., a 48% reduction compared to DiT and a 35% reduction compared to Pixart-alpha). Our source code is available at https://github.com/360CVGroup/Qihoo-T2X.
Authors: Weijie He, Mushui Liu, Yunlong Yu, Zheming Lu, Xi Li
Abstract: Single-frame infrared small target (SIRST) detection poses a significant challenge due to the requirement to discern minute targets amidst complex infrared background clutter. Recently, deep learning approaches have shown promising results in this domain. However, these methods heavily rely on extensive manual annotations, which are particularly cumbersome and resource-intensive for infrared small targets owing to their minute sizes. To address this limitation, we introduce a Hybrid Mask Generation (HMG) approach that recovers high-quality masks for each target from only a single-point label for network training. Specifically, our HMG approach consists of a handcrafted Points-to-Mask Generation strategy coupled with a pseudo mask updating strategy to recover and refine pseudo masks from point labels. The Points-to-Mask Generation strategy divides two distinct stages: Points-to-Box conversion, where individual point labels are transformed into bounding boxes, and subsequently, Box-to-Mask prediction, where these bounding boxes are elaborated into precise masks. The mask updating strategy integrates the complementary strengths of handcrafted and deep-learning algorithms to iteratively refine the initial pseudo masks. Experimental results across three datasets demonstrate that our method outperforms the existing methods for infrared small target detection with single-point supervision.
Authors: Yujun Huang, Bin Chen, Niu Lian, Baoyi An, Shu-Tao Xia
Abstract: Multi-view image compression is vital for 3D-related applications. To effectively model correlations between views, existing methods typically predict disparity between two views on a 2D plane, which works well for small disparities, such as in stereo images, but struggles with larger disparities caused by significant view changes. To address this, we propose a novel approach: learning-based multi-view image coding with 3D Gaussian geometric priors (3D-GP-LMVIC). Our method leverages 3D Gaussian Splatting to derive geometric priors of the 3D scene, enabling more accurate disparity estimation across views within the compression model. Additionally, we introduce a depth map compression model to reduce redundancy in geometric information between views. A multi-view sequence ordering method is also proposed to enhance correlations between adjacent views. Experimental results demonstrate that 3D-GP-LMVIC surpasses both traditional and learning-based methods in performance, while maintaining fast encoding and decoding speed.
Authors: Boyuan Tian, Yihan Pang, Muhammad Huzaifa, Shenlong Wang, Sarita Adve
Abstract: Extended Reality (XR) enables immersive experiences through untethered headsets but suffers from stringent battery and resource constraints. Energy-efficient design is crucial to ensure both longevity and high performance in XR devices. However, latency and accuracy are often prioritized over energy, leading to a gap in achieving energy efficiency. This paper examines scene reconstruction, a key building block for immersive XR experiences, and demonstrates how energy efficiency can be achieved by navigating the trilemma of energy, latency, and accuracy. We explore three classes of energy-oriented optimizations, covering the algorithm, execution, and data, that reveal a broad design space through configurable parameters. Our resulting 72 designs expose a wide range of latency and energy trade-offs, with a smaller range of accuracy loss. We identify a Pareto-optimal curve and show that the designs on the curve are achievable only through synergistic co-optimization of all three optimization classes and by considering the latency and accuracy needs of downstream scene reconstruction consumers. Our analysis covering various use cases and measurements on an embedded class system shows that, relative to the baseline, our designs offer energy benefits of up to 60X with potential latency range of 4X slowdown to 2X speedup. Detailed exploration of a use case across representative data sequences from ScanNet showed about 25X energy savings with 1.5X latency reduction and negligible reconstruction quality loss.
Authors: Yangguang Chen, Tong Wang, Guanzhou Chen, Kun Zhu, Xiaoliang Tan, Jiaqi Wang, Hong Xie, Wenlin Zhou, Jingyi Zhao, Qing Wang, Xiaolong Luo, Xiaodong Zhang
Abstract: Detection of building facade attachments such as doors, windows, balconies, air conditioner units, billboards, and glass curtain walls plays a pivotal role in numerous applications. Building facade attachments detection aids in vbuilding information modeling (BIM) construction and meeting Level of Detail 3 (LOD3) standards. Yet, it faces challenges like uneven object distribution, small object detection difficulty, and background interference. To counter these, we propose BFA-YOLO, a model for detecting facade attachments in multi-view images. BFA-YOLO incorporates three novel innovations: the Feature Balanced Spindle Module (FBSM) for addressing uneven distribution, the Target Dynamic Alignment Task Detection Head (TDATH) aimed at improving small object detection, and the Position Memory Enhanced Self-Attention Mechanism (PMESA) to combat background interference, with each component specifically designed to solve its corresponding challenge. Detection efficacy of deep network models deeply depends on the dataset's characteristics. Existing open source datasets related to building facades are limited by their single perspective, small image pool, and incomplete category coverage. We propose a novel method for building facade attachments detection dataset construction and construct the BFA-3D dataset for facade attachments detection. The BFA-3D dataset features multi-view, accurate labels, diverse categories, and detailed classification. BFA-YOLO surpasses YOLOv8 by 1.8% and 2.9% in mAP@0.5 on the multi-view BFA-3D and street-view Facade-WHU datasets, respectively. These results underscore BFA-YOLO's superior performance in detecting facade attachments.
Authors: Woojin Cho, Jihyun Lee, Minjae Yi, Minje Kim, Taeyun Woo, Donghwan Kim, Taewook Ha, Hyokeun Lee, Je-Hwan Ryu, Woontack Woo, Tae-Kyun Kim
Abstract: Existing datasets for 3D hand-object interaction are limited either in the data cardinality, data variations in interaction scenarios, or the quality of annotations. In this work, we present a comprehensive new training dataset for hand-object interaction called HOGraspNet. It is the only real dataset that captures full grasp taxonomies, providing grasp annotation and wide intraclass variations. Using grasp taxonomies as atomic actions, their space and time combinatorial can represent complex hand activities around objects. We select 22 rigid objects from the YCB dataset and 8 other compound objects using shape and size taxonomies, ensuring coverage of all hand grasp configurations. The dataset includes diverse hand shapes from 99 participants aged 10 to 74, continuous video frames, and a 1.5M RGB-Depth of sparse frames with annotations. It offers labels for 3D hand and object meshes, 3D keypoints, contact maps, and \emph{grasp labels}. Accurate hand and object 3D meshes are obtained by fitting the hand parametric model (MANO) and the hand implicit function (HALO) to multi-view RGBD frames, with the MoCap system only for objects. Note that HALO fitting does not require any parameter tuning, enabling scalability to the dataset's size with comparable accuracy to MANO. We evaluate HOGraspNet on relevant tasks: grasp classification and 3D hand pose estimation. The result shows performance variations based on grasp type and object class, indicating the potential importance of the interaction space captured by our dataset. The provided data aims at learning universal shape priors or foundation models for 3D hand-object interaction. Our dataset and code are available at https://hograspnet2024.github.io/.
Authors: Yin Tang, Wei Luo, Jinrui Zhang, Wei Huang, Ruihai Jing, Deyu Zhang
Abstract: Multi-instance Repetitive Action Counting (MRAC) aims to estimate the number of repetitive actions performed by multiple instances in untrimmed videos, commonly found in human-centric domains like sports and exercise. In this paper, we propose MultiCounter, a fully end-to-end deep learning framework that enables simultaneous detection, tracking, and counting of repetitive actions of multiple human instances. Specifically, MultiCounter incorporates two novel modules: 1) mixed spatiotemporal interaction for efficient context correlation across consecutive frames, and 2) task-specific heads for accurate perception of periodic boundaries and generalization for action-agnostic human instances. We train MultiCounter on a synthetic dataset called MultiRep generated from annotated real-world videos. Experiments on the MultiRep dataset validate the fundamental challenge of MRAC tasks and showcase the superiority of our proposed model. Compared to ByteTrack+RepNet, a solution that combines an advanced tracker with a single repetition counter, MultiCounter substantially improves Period-mAP by 41.0%, reduces AvgMAE by 58.6%, and increases AvgOBO 1.48 times. This sets a new benchmark in the field of MRAC. Moreover, MultiCounter runs in real-time on a commodity GPU server and is insensitive to the number of human instances in a video.
Authors: Tianqi Wei, Zhi Chen, Xin Yu, Scott Chapman, Paul Melloy, Zi Huang
Abstract: Plant diseases pose significant threats to agriculture. It necessitates proper diagnosis and effective treatment to safeguard crop yields. To automate the diagnosis process, image segmentation is usually adopted for precisely identifying diseased regions, thereby advancing precision agriculture. Developing robust image segmentation models for plant diseases demands high-quality annotations across numerous images. However, existing plant disease datasets typically lack segmentation labels and are often confined to controlled laboratory settings, which do not adequately reflect the complexity of natural environments. Motivated by this fact, we established PlantSeg, a large-scale segmentation dataset for plant diseases. PlantSeg distinguishes itself from existing datasets in three key aspects. (1) Annotation type: Unlike the majority of existing datasets that only contain class labels or bounding boxes, each image in PlantSeg includes detailed and high-quality segmentation masks, associated with plant types and disease names. (2) Image source: Unlike typical datasets that contain images from laboratory settings, PlantSeg primarily comprises in-the-wild plant disease images. This choice enhances the practical applicability, as the trained models can be applied for integrated disease management. (3) Scale: PlantSeg is extensive, featuring 11,400 images with disease segmentation masks and an additional 8,000 healthy plant images categorized by plant type. Extensive technical experiments validate the high quality of PlantSeg's annotations. This dataset not only allows researchers to evaluate their image classification methods but also provides a critical foundation for developing and benchmarking advanced plant disease segmentation algorithms.
Authors: Rahul Ramachandran, Tejal Kulkarni, Charchit Sharma, Deepak Vijaykeerthy, Vineeth N Balasubramanian
Abstract: Evaluating models and datasets in computer vision remains a challenging task, with most leaderboards relying solely on accuracy. While accuracy is a popular metric for model evaluation, it provides only a coarse assessment by considering a single model's score on all dataset items. This paper explores Item Response Theory (IRT), a framework that infers interpretable latent parameters for an ensemble of models and each dataset item, enabling richer evaluation and analysis beyond the single accuracy number. Leveraging IRT, we assess model calibration, select informative data subsets, and demonstrate the usefulness of its latent parameters for analyzing and comparing models and datasets in computer vision.
Authors: Koen Kraaijveld, Yifan Jiang, Kaixin Ma, Filip Ilievski
Abstract: While visual question-answering (VQA) benchmarks have catalyzed the development of reasoning techniques, they have focused on vertical thinking. Effective problem-solving also necessitates lateral thinking, which remains understudied in AI and has not been used to test visual perception systems. To bridge this gap, we formulate visual lateral thinking as a multiple-choice question-answering task and describe a three-step taxonomy-driven methodology for instantiating task examples. Then, we develop COLUMBUS, a synthetic benchmark that applies the task pipeline to create QA sets with text and icon rebus puzzles based on publicly available collections of compounds and common phrases. COLUMBUS comprises over 1,000 puzzles, each with four answer candidates. While the SotA vision-language models (VLMs) achieve decent performance, our evaluation demonstrates a substantial gap between humans and models. VLMs benefit from human-curated descriptions but struggle to self-generate such representations at the right level of abstraction.
Authors: Kentaro Hirahara, Chikahito Nakane, Hajime Ebisawa, Tsuyoshi Kuroda, Yohei Iwaki, Tomoyoshi Utsumi, Yuichiro Nomura, Makoto Koike, Hiroshi Mineno
Abstract: In an agricultural field, plant phenotyping using object detection models is gaining attention. However, collecting the training data necessary to create generic and high-precision models is extremely challenging due to the difficulty of annotation and the diversity of domains. Furthermore, it is difficult to transfer training data across different crops, and although machine learning models effective for specific environments, conditions, or crops have been developed, they cannot be widely applied in actual fields. In this study, we propose a generative data augmentation method (D4) for vineyard shoot detection. D4 uses a pre-trained text-guided diffusion model based on a large number of original images culled from video data collected by unmanned ground vehicles or other means, and a small number of annotated datasets. The proposed method generates new annotated images with background information adapted to the target domain while retaining annotation information necessary for object detection. In addition, D4 overcomes the lack of training data in agriculture, including the difficulty of annotation and diversity of domains. We confirmed that this generative data augmentation method improved the mean average precision by up to 28.65% for the BBox detection task and the average precision by up to 13.73% for the keypoint detection task for vineyard shoot detection. Our generative data augmentation method D4 is expected to simultaneously solve the cost and domain diversity issues of training data generation in agriculture and improve the generalization performance of detection models.
Authors: Shu-Min Tan, Shih-Hsun Hung, Je-Chiang Tsai
Abstract: Coffee is one of the most valuable primary commodities. Despite this, the common selection technique of green coffee beans relies on personnel visual inspection, which is labor-intensive and subjective. Therefore, an efficient way to evaluate the quality of beans is needed. In this paper, we demonstrate a site-independent approach to find site-specific color features of the seed coat in qualified green coffee beans. We then propose two evaluation schemes for green coffee beans based on this site-specific color feature of qualified beans. Due to the site-specific properties of these color features, machine learning classifiers indicate that compared with the existing evaluation schemes of beans, our evaluation schemes have the advantages of being simple, having less computational costs, and having universal applicability. Finally, this site-specific color feature can distinguish qualified beans from different growing sites. Moreover, this function can prevent cheating in the coffee business and is unique to our evaluation scheme of beans.
Authors: Yi Tian, Juan Andrade-Cetto
Abstract: Event cameras generate asynchronous and sparse event streams capturing changes in light intensity. They offer significant advantages over conventional frame-based cameras, such as a higher dynamic range and an extremely faster data rate, making them particularly useful in scenarios involving fast motion or challenging lighting conditions. Spiking neural networks (SNNs) share similar asynchronous and sparse characteristics and are well-suited for processing data from event cameras. Inspired by the potential of transformers and spike-driven transformers (spikeformers) in other computer vision tasks, we propose two solutions for fast and robust optical flow estimation for event cameras: STTFlowNet and SDformerFlow. STTFlowNet adopts a U-shaped artificial neural network (ANN) architecture with spatiotemporal shifted window self-attention (swin) transformer encoders, while SDformerFlow presents its fully spiking counterpart, incorporating swin spikeformer encoders. Furthermore, we present two variants of the spiking version with different neuron models. Our work is the first to make use of spikeformers for dense optical flow estimation. We conduct end-to-end training for all models using supervised learning. Our results yield state-of-the-art performance among SNN-based event optical flow methods on both the DSEC and MVSEC datasets, and show significant reduction in power consumption compared to the equivalent ANNs.
Authors: Tim Bader, Leon Eisemann, Adrian Pogorzelski, Namrata Jangid, Attila-Balazs Kis
Abstract: The increasing accuracy reports of metric monocular depth estimation models lead to a growing interest from the automotive domain. Current model evaluations do not provide deeper insights into the models' performance, also in relation to safety-critical or unseen classes. Within this paper, we present a novel approach for the evaluation of depth estimation models. Our proposed metric leverages three components, a class-wise component, an edge and corner image feature component, and a global consistency retaining component. Classes are further weighted on their distance in the scene and on criticality for automotive applications. In the evaluation, we present the benefits of our metric through comparison to classical metrics, class-wise analytics, and the retrieval of critical situations. The results show that our metric provides deeper insights into model results while fulfilling safety-critical requirements. We release the code and weights on the following repository: \href{https://github.com/leisemann/ca_mmde}
Authors: Yi Zhu, Yanpeng Zhou, Chunwei Wang, Yang Cao, Jianhua Han, Lu Hou, Hang Xu
Abstract: Currently, vision encoder models like Vision Transformers (ViTs) typically excel at image recognition tasks but cannot simultaneously support text recognition like human visual recognition. To address this limitation, we propose UNIT, a novel training framework aimed at UNifying Image and Text recognition within a single model. Starting with a vision encoder pre-trained with image recognition tasks, UNIT introduces a lightweight language decoder for predicting text outputs and a lightweight vision decoder to prevent catastrophic forgetting of the original image encoding capabilities. The training process comprises two stages: intra-scale pretraining and inter-scale finetuning. During intra-scale pretraining, UNIT learns unified representations from multi-scale inputs, where images and documents are at their commonly used resolution, to enable fundamental recognition capability. In the inter-scale finetuning stage, the model introduces scale-exchanged data, featuring images and documents at resolutions different from the most commonly used ones, to enhance its scale robustness. Notably, UNIT retains the original vision encoder architecture, making it cost-free in terms of inference and deployment. Experiments across multiple benchmarks confirm that our method significantly outperforms existing methods on document-related tasks (e.g., OCR and DocQA) while maintaining the performances on natural images, demonstrating its ability to substantially enhance text recognition without compromising its core image recognition capabilities.
Authors: Gustav Grund Pihlgren, Kary Fr\"amling
Abstract: Perturbation-based post-hoc image explanation methods are commonly used to explain image prediction models by perturbing parts of the input to measure how those parts affect the output. Due to the intractability of perturbing each pixel individually, images are typically attributed to larger segments. The Randomized Input Sampling for Explanations (RISE) method solved this issue by using smooth perturbation masks. While this method has proven effective and popular, it has not been investigated which parts of the method are responsible for its success. This work tests many combinations of mask sampling, segmentation techniques, smoothing, and attribution calculation. The results show that the RISE-style pixel attribution is beneficial to all evaluated methods. Furthermore, it is shown that attribution calculation is the least impactful parameter. The implementation of this work is available online: https://github.com/guspih/post-hoc-image-perturbation.
URLs: https://github.com/guspih/post-hoc-image-perturbation.
Authors: Arthur Hemmer, Micka\"el Coustaty, Nicola Bartolo, Jean-Marc Ogier
Abstract: Optical Character Recognition (OCR) continues to face accuracy challenges that impact subsequent applications. To address these errors, we explore the utility of OCR confidence scores for enhancing post-OCR error detection. Our study involves analyzing the correlation between confidence scores and error rates across different OCR systems. We develop ConfBERT, a BERT-based model that incorporates OCR confidence scores into token embeddings and offers an optional pre-training phase for noise adjustment. Our experimental results demonstrate that integrating OCR confidence scores can enhance error detection capabilities. This work underscores the importance of OCR confidence scores in improving detection accuracy and reveals substantial disparities in performance between commercial and open-source OCR technologies.
Authors: Hangcheng Cao, Longzhi Yuan, Guowen Xu, Ziyang He, Zhengru Fang, Yuguang Fang
Abstract: Traffic sign recognition systems play a crucial role in assisting drivers to make informed decisions while driving. However, due to the heavy reliance on deep learning technologies, particularly for future connected and autonomous driving, these systems are susceptible to adversarial attacks that pose significant safety risks to both personal and public transportation. Notably, researchers recently identified a new attack vector to deceive sign recognition systems: projecting well-designed adversarial light patches onto traffic signs. In comparison with traditional adversarial stickers or graffiti, these emerging light patches exhibit heightened aggression due to their ease of implementation and outstanding stealthiness. To effectively counter this security threat, we propose a universal image inpainting mechanism, namely, SafeSign. It relies on attention-enabled multi-view image fusion to repair traffic signs contaminated by adversarial light patches, thereby ensuring the accurate sign recognition. Here, we initially explore the fundamental impact of malicious light patches on the local and global feature spaces of authentic traffic signs. Then, we design a binary mask-based U-Net image generation pipeline outputting diverse contaminated sign patterns, to provide our image inpainting model with needed training data. Following this, we develop an attention mechanism-enabled neural network to jointly utilize the complementary information from multi-view images to repair contaminated signs. Finally, extensive experiments are conducted to evaluate SafeSign's effectiveness in resisting potential light patch-based attacks, bringing an average accuracy improvement of 54.8% in three widely-used sign recognition models
Authors: Ting-Ru Liu, Hsuan-Kung Yang, Jou-Min Liu, Chun-Wei Huang, Tsung-Chih Chiang, Quan Kong, Norimasa Kobori, Chun-Yi Lee
Abstract: Scene coordinate regression (SCR) methods have emerged as a promising area of research due to their potential for accurate visual localization. However, many existing SCR approaches train on samples from all image regions, including dynamic objects and texture-less areas. Utilizing these areas for optimization during training can potentially hamper the overall performance and efficiency of the model. In this study, we first perform an in-depth analysis to validate the adverse impacts of these areas. Drawing inspiration from our analysis, we then introduce an error-guided feature selection (EGFS) mechanism, in tandem with the use of the Segment Anything Model (SAM). This mechanism seeds low reprojection areas as prompts and expands them into error-guided masks, and then utilizes these masks to sample points and filter out problematic areas in an iterative manner. The experiments demonstrate that our method outperforms existing SCR approaches that do not rely on 3D information on the Cambridge Landmarks and Indoor6 datasets.
Authors: Jumabek Alikhanov, Dilshod Obidov, Hakil Kim
Abstract: The Lightweight Integrated Tracking-Feature Extraction (LITE) paradigm is introduced as a novel multi-object tracking (MOT) approach. It enhances ReID-based trackers by eliminating inference, pre-processing, post-processing, and ReID model training costs. LITE uses real-time appearance features without compromising speed. By integrating appearance feature extraction directly into the tracking pipeline using standard CNN-based detectors such as YOLOv8m, LITE demonstrates significant performance improvements. The simplest implementation of LITE on top of classic DeepSORT achieves a HOTA score of 43.03% at 28.3 FPS on the MOT17 benchmark, making it twice as fast as DeepSORT on MOT17 and four times faster on the more crowded MOT20 dataset, while maintaining similar accuracy. Additionally, a new evaluation framework for tracking-by-detection approaches reveals that conventional trackers like DeepSORT remain competitive with modern state-of-the-art trackers when evaluated under fair conditions. The code will be available post-publication at https://github.com/Jumabek/LITE.
Authors: Lorenza Prospero, Abdullah Hamdi, Joao F. Henriques, Christian Rupprecht
Abstract: Reconstructing realistic 3D human models from monocular images has significant applications in creative industries, human-computer interfaces, and healthcare. We base our work on 3D Gaussian Splatting (3DGS), a scene representation composed of a mixture of Gaussians. Predicting such mixtures for a human from a single input image is challenging, as it is a non-uniform density (with a many-to-one relationship with input pixels) with strict physical constraints. At the same time, it needs to be flexible to accommodate a variety of clothes and poses. Our key observation is that the vertices of standardized human meshes (such as SMPL) can provide an adequate density and approximate initial position for Gaussians. We can then train a transformer model to jointly predict comparatively small adjustments to these positions, as well as the other Gaussians' attributes and the SMPL parameters. We show empirically that this combination (using only multi-view supervision) can achieve fast inference of 3D human models from a single image without test-time optimization, expensive diffusion models, or 3D points supervision. We also show that it can improve 3D pose estimation by better fitting human models that account for clothes and other variations. The code is available on the project website https://abdullahamdi.com/gst/ .
Authors: Aglind Reka, Diana Laura Borza, Dominick Reilly, Michal Balazia, Francois Bremond
Abstract: Temporal Action Detection (TAD), the task of localizing and classifying actions in untrimmed video, remains challenging due to action overlaps and variable action durations. Recent findings suggest that TAD performance is dependent on the structural design of transformers rather than on the self-attention mechanism. Building on this insight, we propose a refined feature extraction process through lightweight, yet effective operations. First, we employ a local branch that employs parallel convolutions with varying window sizes to capture both fine-grained and coarse-grained temporal features. This branch incorporates a gating mechanism to select the most relevant features. Second, we introduce a context branch that uses boundary frames as key-value pairs to analyze their relationship with the central frame through cross-attention. The proposed method captures temporal dependencies and improves contextual understanding. Evaluations of the gating mechanism and context branch on challenging datasets (THUMOS14 and EPIC-KITCHEN 100) show a consistent improvement over the baseline and existing methods.
Authors: Yuan Bian, Min Liu, Xueping Wang, Yunfeng Ma, Yaonan Wang
Abstract: Deep learning-based person re-identification (re-id) models are widely employed in surveillance systems and inevitably inherit the vulnerability of deep networks to adversarial attacks. Existing attacks merely consider cross-dataset and cross-model transferability, ignoring the cross-test capability to perturb models trained in different domains. To powerfully examine the robustness of real-world re-id models, the Meta Transferable Generative Attack (MTGA) method is proposed, which adopts meta-learning optimization to promote the generative attacker producing highly transferable adversarial examples by learning comprehensively simulated transfer-based cross-model\&dataset\&test black-box meta attack tasks. Specifically, cross-model\&dataset black-box attack tasks are first mimicked by selecting different re-id models and datasets for meta-train and meta-test attack processes. As different models may focus on different feature regions, the Perturbation Random Erasing module is further devised to prevent the attacker from learning to only corrupt model-specific features. To boost the attacker learning to possess cross-test transferability, the Normalization Mix strategy is introduced to imitate diverse feature embedding spaces by mixing multi-domain statistics of target models. Extensive experiments show the superiority of MTGA, especially in cross-model\&dataset and cross-model\&dataset\&test attacks, our MTGA outperforms the SOTA methods by 21.5\% and 11.3\% on mean mAP drop rate, respectively. The code of MTGA will be released after the paper is accepted.
Authors: Zeren Zhang, Jo-Ku Cheng, Jingyang Deng, Lu Tian, Jinwen Ma, Ziran Qin, Xiaokai Zhang, Na Zhu, Tuo Leng
Abstract: Mathematical reasoning remains an ongoing challenge for AI models, especially for geometry problems that require both linguistic and visual signals. As the vision encoders of most MLLMs are trained on natural scenes, they often struggle to understand geometric diagrams, performing no better in geometry problem solving than LLMs that only process text. This limitation is amplified by the lack of effective methods for representing geometric relationships. To address these issues, we introduce the Diagram Formalization Enhanced Geometry Problem Solver (DFE-GPS), a new framework that integrates visual features, geometric formal language, and natural language representations. We propose a novel synthetic data approach and create a large-scale geometric dataset, SynthGeo228K, annotated with both formal and natural language captions, designed to enhance the vision encoder for a better understanding of geometric structures. Our framework improves MLLMs' ability to process geometric diagrams and extends their application to open-ended tasks on the formalgeo7k dataset.
Authors: Yubiao Yue, Jun Xue, Haihuang Liang, Zhenzhang Li, Yufeng Wang
Abstract: Due to the lack of effective mpox detection tools, the mpox virus continues to spread worldwide and has once again been declared a public health emergency of international concern by the World Health Organization. Deep learning-based mpox detection tools are crucial to alleviate mpox outbreak. However, existing methods have difficulty in achieving a good trade-off between detection performance, parameter size, and model complexity, which is crucial for practical applications and widespread deployment, especially in resource-limited scenarios. Given that the success of Mamba in modeling long-range dependencies and its linear complexity, we proposed a lightweight hybrid architecture called MpoxMamba. MpoxMamba utilizes deep separable convolutions to extract local feature representations in mpox skin lesions, and greatly enhances the model's ability to model the global contextual information by grouped Mamba modules. Experimental results on two widely recognized mpox datasets demonstrate that MpoxMamba outperforms existing mpox detection methods and state-of-the-art lightweight models. We also developed a web-based online application to provide free mpox detection services to the public in the epidemic areas (http://5227i971s5.goho.co:30290). The source codes of MpoxMamba are available at https://github.com/YubiaoYue/MpoxMamba.
URLs: http://5227i971s5.goho.co:30290)., https://github.com/YubiaoYue/MpoxMamba.
Authors: Maksim Kolodiazhnyi, Anna Vorontsova, Matvey Skripkin, Danila Rukhovich, Anton Konushin
Abstract: Growing customer demand for smart solutions in robotics and augmented reality has attracted considerable attention to 3D object detection from point clouds. Yet, existing indoor datasets taken individually are too small and insufficiently diverse to train a powerful and general 3D object detection model. In the meantime, more general approaches utilizing foundation models are still inferior in quality to those based on supervised training for a specific task. In this work, we propose \ours{}, a simple yet effective 3D object detection model, which is trained on a mixture of indoor datasets and is capable of working in various indoor environments. By unifying different label spaces, \ours{} enables learning a strong representation across multiple datasets through a supervised joint training scheme. The proposed network architecture is built upon a vanilla transformer encoder, making it easy to run, customize and extend the prediction pipeline for practical use. Extensive experiments demonstrate that \ours{} obtains significant gains over existing 3D object detection methods in 6 indoor benchmarks: ScanNet (+1.1 mAP50), ARKitScenes (+19.4 mAP25), S3DIS (+9.1 mAP50), MultiScan (+9.3 mAP50), 3RScan (+3.2 mAP50), and ScanNet++ (+2.7 mAP50). Code is available at https://github.com/filapro/unidet3d .
Authors: Yang Zhao, Gangwei Xu, Gang Wu
Abstract: Current state-of-the-art flow methods are mostly based on dense all-pairs cost volumes. However, as image resolution increases, the computational and spatial complexity of constructing these cost volumes grows at a quartic rate, making these methods impractical for high-resolution images. In this paper, we propose a novel Hybrid Cost Volume for memory-efficient optical flow, named HCV. To construct HCV, we first propose a Top-k strategy to separate the 4D cost volume into two global 3D cost volumes. These volumes significantly reduce memory usage while retaining a substantial amount of matching information. We further introduce a local 4D cost volume with a local search space to supplement the local information for HCV. Based on HCV, we design a memory-efficient optical flow network, named HCVFlow. Compared to the recurrent flow methods based the all-pairs cost volumes, our HCVFlow significantly reduces memory consumption while ensuring high accuracy. We validate the effectiveness and efficiency of our method on the Sintel and KITTI datasets and real-world 4K (2160*3840) resolution images. Extensive experiments show that our HCVFlow has very low memory usage and outperforms other memory-efficient methods in terms of accuracy. The code is publicly available at https://github.com/gangweiX/HCVFlow.
Authors: Changsong Liu, Wei Zhang, Yanyan Liu, Mingyang Li, Wenlin Li, Yimeng Fan, Xiangnan Bai, Liang Zhangd
Abstract: Edge detection, as a fundamental task in computer vision, has garnered increasing attention. The advent of deep learning has significantly advanced this field. However, recent deep learning-based methods which rely on large-scale pre-trained weights cannot be trained from scratch, with very limited research addressing this issue. This paper proposes a novel cycle pixel difference convolution (CPDC), which effectively integrates image gradient information with modern convolution operations. Based on the CPDC, we develop a U-shape encoder-decoder model named CPD-Net, which is a purely end-to-end network. Additionally, to address the issue of edge thickness produced by most existing methods, we construct a multi-scale information enhancement module (MSEM) to enhance the discriminative ability of the model, thereby generating crisp and clean contour maps. Comprehensive experiments conducted on three standard benchmarks demonstrate that our method achieves competitive performance on the BSDS500 dataset (ODS=0.813), NYUD-V2 (ODS=0.760), and BIPED dataset (ODS=0.898). Our approach provides a novel perspective for addressing these challenges in edge detection.
Authors: Yunhao Bai, Qinji Yu, Boxiang Yun, Dakai Jin, Yingda Xia, Yan Wang
Abstract: The Segment Anything Model 2 (SAM2) has recently demonstrated exceptional performance in zero-shot prompt segmentation for natural images and videos. However, it faces significant challenges when applied to medical images. Since its release, many attempts have been made to adapt SAM2's segmentation capabilities to the medical imaging domain. These efforts typically involve using a substantial amount of labeled data to fine-tune the model's weights. In this paper, we explore SAM2 from a different perspective via making the full use of its trained memory attention module and its ability of processing mask prompts. We introduce FS-MedSAM2, a simple yet effective framework that enables SAM2 to achieve superior medical image segmentation in a few-shot setting, without the need for fine-tuning. Our framework outperforms the current state-of-the-arts on two publicly available medical image datasets. The code is available at https://github.com/DeepMed-Lab-ECNU/FS_MedSAM2.
Authors: Bappaditya Dey, Matthias Monden, Victor Blanco, Sandip Halder, Stefan De Gendt
Abstract: In this research, we introduce a unified end-to-end Automated Defect Classification-Detection-Segmentation (ADCDS) framework for classifying, detecting, and segmenting multiple instances of semiconductor defects for advanced nodes. This framework consists of two modules: (a) a defect detection module, followed by (b) a defect segmentation module. The defect detection module employs Deformable DETR to aid in the classification and detection of nano-scale defects, while the segmentation module utilizes BoxSnake. BoxSnake facilitates box-supervised instance segmentation of nano-scale defects, supported by the former module. This simplifies the process by eliminating the laborious requirement for ground-truth pixel-wise mask annotation by human experts, which is typically associated with training conventional segmentation models. We have evaluated the performance of our ADCDS framework using two distinct process datasets from real wafers, as ADI and AEI, specifically focusing on Line-space patterns. We have demonstrated the applicability and significance of our proposed methodology, particularly in the nano-scale segmentation and generation of binary defect masks, using the challenging ADI SEM dataset where ground-truth pixelwise segmentation annotations were unavailable. Furthermore, we have presented a comparative analysis of our proposed framework against previous approaches to demonstrate its effectiveness. Our proposed framework achieved an overall mAP@IoU0.5 of 72.19 for detection and 78.86 for segmentation on the ADI dataset. Similarly, for the AEI dataset, these metrics were 90.38 for detection and 95.48 for segmentation. Thus, our proposed framework effectively fulfils the requirements of advanced defect analysis while addressing significant constraints.
Authors: Isabela Borlido, Eduardo Bouhid, Victor Sundermann, Hugo Resende, Alvaro Luiz Fazenda, Fabio Faria, Silvio Jamil F. Guimar\~aes
Abstract: The conservation of tropical forests is a topic of significant social and ecological relevance due to their crucial role in the global ecosystem. Unfortunately, deforestation and degradation impact millions of hectares annually, requiring government or private initiatives for effective forest monitoring. However, identifying deforested regions in satellite images is challenging due to data imbalance, image resolution, low-contrast regions, and occlusion. Superpixel segmentation can overcome these drawbacks, reducing workload and preserving important image boundaries. However, most works for remote sensing images do not exploit recent superpixel methods. In this work, we evaluate 16 superpixel methods in satellite images to support a deforestation detection system in tropical forests. We also assess the performance of superpixel methods for the target task, establishing a relationship with segmentation methodological evaluation. According to our results, ERS, GMMSP, and DISF perform best on UE, BR, and SIRS, respectively, whereas ERS has the best trade-off with CO and Reg. In classification, SH, DISF, and ISF perform best on RGB, UMDA, and PCA compositions, respectively. According to our experiments, superpixel methods with better trade-offs between delineation, homogeneity, compactness, and regularity are more suitable for identifying good superpixels for deforestation detection tasks.
Authors: Ryan A. Subong, Alma Jean D. Subong
Abstract: This paper aims to verify the effectiveness of the software implementation of the proposed algorithm in creating computer-generated images of sand mixtures using a photograph of sand as an input and its effectiveness in converting digital pictures into sand-based images out of the mixtures it generated. The method of this paper is to visually compare the photographed image of the actual mixtures to its computer-generated counterpart to verify if the mixture generation produces results as expected and compare the computer-generated sand-based images with its source to verify image reproduction maintains same image content. The results of the mixture comparison shows that the actual and the computer-generated ones have similar overall shade and color. Still, the generated one has a rougher texture and higher contrast due to the method of inheriting visual features by pixel, not by individual sand particles. The comparison of the sand-based image and its source has demonstrated the software's ability to maintain the essence of its contents during conversion while replacing its texture with the visual properties of the generated sand mixture. The result have shown that the software implementation of the proposed algorithm can effectively use the images of sand to generate images of its mixtures and use those mixture images to convert a digital picture into a computer-generated sand-based image.
Authors: Hongqiu Wang, Yixian Chen, Wu Chen, Huihui Xu, Haoyu Zhao, Bin Sheng, Huazhu Fu, Guang Yang, Lei Zhu
Abstract: Ultra-Wide-Field Scanning Laser Ophthalmoscopy (UWF-SLO) images capture high-resolution views of the retina with typically 200 spanning degrees. Accurate segmentation of vessels in UWF-SLO images is essential for detecting and diagnosing fundus disease. Recent studies have revealed that the selective State Space Model (SSM) in Mamba performs well in modeling long-range dependencies, which is crucial for capturing the continuity of elongated vessel structures. Inspired by this, we propose the first Serpentine Mamba (Serp-Mamba) network to address this challenging task. Specifically, we recognize the intricate, varied, and delicate nature of the tubular structure of vessels. Furthermore, the high-resolution of UWF-SLO images exacerbates the imbalance between the vessel and background categories. Based on the above observations, we first devise a Serpentine Interwoven Adaptive (SIA) scan mechanism, which scans UWF-SLO images along curved vessel structures in a snake-like crawling manner. This approach, consistent with vascular texture transformations, ensures the effective and continuous capture of curved vascular structure features. Second, we propose an Ambiguity-Driven Dual Recalibration (ADDR) module to address the category imbalance problem intensified by high-resolution images. Our ADDR module delineates pixels by two learnable thresholds and refines ambiguous pixels through a dual-driven strategy, thereby accurately distinguishing vessels and background regions. Experiment results on three datasets demonstrate the superior performance of our Serp-Mamba on high-resolution vessel segmentation. We also conduct a series of ablation studies to verify the impact of our designs. Our code shall be released upon publication of this work.
Authors: Gianluca Carloni, Sara Colantonio
Abstract: The aim of this paper is threefold. We inform the AI practitioner about the human visual system with an extensive literature review; we propose a novel biologically motivated neural network for image classification; and, finally, we present a new plug-and-play module to model context awareness. We focus on the effect of incorporating circuit motifs found in biological brains to address visual recognition. Our convolutional architecture is inspired by the connectivity of human cortical and subcortical streams, and we implement bottom-up and top-down modulations that mimic the extensive afferent and efferent connections between visual and cognitive areas. Our Contextual Attention Block is simple and effective and can be integrated with any feed-forward neural network. It infers weights that multiply the feature maps according to their causal influence on the scene, modeling the co-occurrence of different objects in the image. We place our module at different bottlenecks to infuse a hierarchical context awareness into the model. We validated our proposals through image classification experiments on benchmark data and found a consistent improvement in performance and the robustness of the produced explanations via class activation. Our code is available at https://github.com/gianlucarloni/CoCoReco.
Authors: Hao Luo, Baoliang Chen, Lingyu Zhu, Peilin Chen, Shiqi Wang
Abstract: Scene observation from multiple perspectives would bring a more comprehensive visual experience. However, in the context of acquiring multiple views in the dark, the highly correlated views are seriously alienated, making it challenging to improve scene understanding with auxiliary views. Recent single image-based enhancement methods may not be able to provide consistently desirable restoration performance for all views due to the ignorance of potential feature correspondence among different views. To alleviate this issue, we make the first attempt to investigate multi-view low-light image enhancement. First, we construct a new dataset called Multi-View Low-light Triplets (MVLT), including 1,860 pairs of triple images with large illumination ranges and wide noise distribution. Each triplet is equipped with three different viewpoints towards the same scene. Second, we propose a deep multi-view enhancement framework based on the Recurrent Collaborative Network (RCNet). Specifically, in order to benefit from similar texture correspondence across different views, we design the recurrent feature enhancement, alignment and fusion (ReEAF) module, in which intra-view feature enhancement (Intra-view EN) followed by inter-view feature alignment and fusion (Inter-view AF) is performed to model the intra-view and inter-view feature propagation sequentially via multi-view collaboration. In addition, two different modules from enhancement to alignment (E2A) and from alignment to enhancement (A2E) are developed to enable the interactions between Intra-view EN and Inter-view AF, which explicitly utilize attentive feature weighting and sampling for enhancement and alignment, respectively. Experimental results demonstrate that our RCNet significantly outperforms other state-of-the-art methods. All of our dataset, code, and model will be available at https://github.com/hluo29/RCNet.
Authors: Xiaoyi Liu, Zhou Yu, Lianghao Tan, Yafeng Yan, Ge Shi
Abstract: Skin lesions are an increasingly significant medical concern, varying widely in severity from benign to cancerous. Accurate diagnosis is essential for ensuring timely and appropriate treatment. This study examines the implementation of deep learning methods to assist in the diagnosis of skin lesions using the HAM10000 dataset, which contains seven distinct types of lesions. First, we evaluated three pre-trained models: MobileNetV2, ResNet18, and VGG11, achieving accuracies of 0.798, 0.802, and 0.805, respectively. To further enhance classification accuracy, we developed ensemble models employing max voting, average voting, and stacking, resulting in accuracies of 0.803, 0.82, and 0.83. Building on the best-performing ensemble learning model, stacking, we developed our proposed model, SkinNet, which incorporates a customized architecture and fine-tuning, achieving an accuracy of 0.867 and an AUC of 0.96. This substantial improvement over individual models demonstrates the effectiveness of ensemble learning in improving skin lesion classification.
Authors: Charlesquin Kemajou Mbakam, Jean-Francois Giovannelli, Marcelo Pereyra
Abstract: Score-based diffusion methods provide a powerful strategy to solve image restoration tasks by flexibly combining a pre-trained foundational prior model with a likelihood function specified during test time. Such methods are predominantly derived from two stochastic processes: reversing Ornstein-Uhlenbeck, which underpins the celebrated denoising diffusion probabilistic models (DDPM) and denoising diffusion implicit models (DDIM), and the Langevin diffusion process. The solutions delivered by DDPM and DDIM are often remarkably realistic, but they are not always consistent with measurements because of likelihood intractability issues and the associated required approximations. Alternatively, using a Langevin process circumvents the intractable likelihood issue, but usually leads to restoration results of inferior quality and longer computing times. This paper presents a novel and highly computationally efficient image restoration method that carefully embeds a foundational DDPM denoiser within an empirical Bayesian Langevin algorithm, which jointly calibrates key model hyper-parameters as it estimates the model's posterior mean. Extensive experimental results on three canonical tasks (image deblurring, super-resolution, and inpainting) demonstrate that the proposed approach improves on state-of-the-art strategies both in image estimation accuracy and computing time.
Authors: Hangyu Qin, Junbin Xiao, Angela Yao
Abstract: Multimodal Large Language Models (MLLMs) have shown excellent performance in question-answering of single-event videos. In this paper, we present question-answering dense video events, a novel task that requires answering and grounding the dense-event questions in long videos, thus challenging MLLMs to faithfully comprehend and reason about multiple events occurring over extended time periods. To facilitate the study, we construct DeVE-QA - a dataset featuring 78K questions about 26K events on 10.6K long videos. We then benchmark and show that existing MLLMs excelling at single-event QA struggle to perform well in DeVE-QA. For improvement, we propose DeVi, a novel training-free MLLM approach that highlights a hierarchical captioning module, a temporal event memory module, and a self-consistency checking module to respectively detect, contextualize and memorize, and ground dense-events in long videos for question answering. Extensive experiments show that DeVi is superior at answering dense-event questions and grounding relevant video moments. Compared with existing MLLMs, it achieves a remarkable increase of 4.1 percent and 3.7 percent for G(round)QA accuracy on DeVE-QA and NExT-GQA respectively.
Authors: Rui Yu, Runkai Zhao, Cong Nie, Heng Wang, HuaiCheng Yan, Meng Wang
Abstract: Accurate and robust LiDAR 3D object detection is essential for comprehensive scene understanding in autonomous driving. Despite its importance, LiDAR detection performance is limited by inherent constraints of point cloud data, particularly under conditions of extended distances and occlusions. Recently, temporal aggregation has been proven to significantly enhance detection accuracy by fusing multi-frame viewpoint information and enriching the spatial representation of objects. In this work, we introduce a novel LiDAR 3D object detection framework, namely LiSTM, to facilitate spatial-temporal feature learning with cross-frame motion forecasting information. We aim to improve the spatial-temporal interpretation capabilities of the LiDAR detector by incorporating a dynamic prior, generated from a non-learnable motion estimation model. Specifically, Motion-Guided Feature Aggregation (MGFA) is proposed to utilize the object trajectory from previous and future motion states to model spatial-temporal correlations into gaussian heatmap over a driving sequence. This motion-based heatmap then guides the temporal feature fusion, enriching the proposed object features. Moreover, we design a Dual Correlation Weighting Module (DCWM) that effectively facilitates the interaction between past and prospective frames through scene- and channel-wise feature abstraction. In the end, a cascade cross-attention-based decoder is employed to refine the 3D prediction. We have conducted experiments on the Waymo and nuScenes datasets to demonstrate that the proposed framework achieves superior 3D detection performance with effective spatial-temporal feature learning.
Authors: Yudi Dai, Zhiyong Wang, Xiping Lin, Chenglu Wen, Lan Xu, Siqi Shen, Yuexin Ma, Cheng Wang
Abstract: We introduce HiSC4D, a novel Human-centered interaction and 4D Scene Capture method, aimed at accurately and efficiently creating a dynamic digital world, containing large-scale indoor-outdoor scenes, diverse human motions, rich human-human interactions, and human-environment interactions. By utilizing body-mounted IMUs and a head-mounted LiDAR, HiSC4D can capture egocentric human motions in unconstrained space without the need for external devices and pre-built maps. This affords great flexibility and accessibility for human-centered interaction and 4D scene capturing in various environments. Taking into account that IMUs can capture human spatially unrestricted poses but are prone to drifting for long-period using, and while LiDAR is stable for global localization but rough for local positions and orientations, HiSC4D employs a joint optimization method, harmonizing all sensors and utilizing environment cues, yielding promising results for long-term capture in large scenes. To promote research of egocentric human interaction in large scenes and facilitate downstream tasks, we also present a dataset, containing 8 sequences in 4 large scenes (200 to 5,000 $m^2$), providing 36k frames of accurate 4D human motions with SMPL annotations and dynamic scenes, 31k frames of cropped human point clouds, and scene mesh of the environment. A variety of scenarios, such as the basketball gym and commercial street, alongside challenging human motions, such as daily greeting, one-on-one basketball playing, and tour guiding, demonstrate the effectiveness and the generalization ability of HiSC4D. The dataset and code will be publicated on www.lidarhumanmotion.net/hisc4d available for research purposes.
Authors: Bj\"orn Michele, Alexandre Boulch, Tuan-Hung Vu, Gilles Puy, Renaud Marlet, Nicolas Courty
Abstract: We tackle the challenging problem of source-free unsupervised domain adaptation (SFUDA) for 3D semantic segmentation. It amounts to performing domain adaptation on an unlabeled target domain without any access to source data; the available information is a model trained to achieve good performance on the source domain. A common issue with existing SFUDA approaches is that performance degrades after some training time, which is a by product of an under-constrained and ill-posed problem. We discuss two strategies to alleviate this issue. First, we propose a sensible way to regularize the learning problem. Second, we introduce a novel criterion based on agreement with a reference model. It is used (1) to stop the training when appropriate and (2) as validator to select hyperparameters without any knowledge on the target domain. Our contributions are easy to implement and readily amenable for all SFUDA methods, ensuring stable improvements over all baselines. We validate our findings on various 3D lidar settings, achieving state-of-the-art performance. The project repository (with code) is: github.com/valeoai/TTYD.
Authors: Zhuoyan Luo, Fengyuan Shi, Yixiao Ge, Yujiu Yang, Limin Wang, Ying Shan
Abstract: We present Open-MAGVIT2, a family of auto-regressive image generation models ranging from 300M to 1.5B. The Open-MAGVIT2 project produces an open-source replication of Google's MAGVIT-v2 tokenizer, a tokenizer with a super-large codebook (i.e., $2^{18}$ codes), and achieves the state-of-the-art reconstruction performance (1.17 rFID) on ImageNet $256 \times 256$. Furthermore, we explore its application in plain auto-regressive models and validate scalability properties. To assist auto-regressive models in predicting with a super-large vocabulary, we factorize it into two sub-vocabulary of different sizes by asymmetric token factorization, and further introduce "next sub-token prediction" to enhance sub-token interaction for better generation quality. We release all models and codes to foster innovation and creativity in the field of auto-regressive visual generation.
Authors: Yecheng Wu, Zhuoyang Zhang, Junyu Chen, Haotian Tang, Dacheng Li, Yunhao Fang, Ligeng Zhu, Enze Xie, Hongxu Yin, Li Yi, Song Han, Yao Lu
Abstract: VILA-U is a Unified foundation model that integrates Video, Image, Language understanding and generation. Traditional visual language models (VLMs) use separate modules for understanding and generating visual content, which can lead to misalignment and increased complexity. In contrast, VILA-U employs a single autoregressive next-token prediction framework for both tasks, eliminating the need for additional components like diffusion models. This approach not only simplifies the model but also achieves near state-of-the-art performance in visual language understanding and generation. The success of VILA-U is attributed to two main factors: the unified vision tower that aligns discrete visual tokens with textual inputs during pretraining, which enhances visual perception, and autoregressive image generation can achieve similar quality as diffusion models with high-quality dataset. This allows VILA-U to perform comparably to more complex models using a fully token-based autoregressive framework.
Authors: Vongani Maluleke, Lea M\"uller, Jathushan Rajasegaran, Georgios Pavlakos, Shiry Ginosar, Angjoo Kanazawa, Jitendra Malik
Abstract: This paper asks to what extent social interaction influences one's behavior. We study this in the setting of two dancers dancing as a couple. We first consider a baseline in which we predict a dancer's future moves conditioned only on their past motion without regard to their partner. We then investigate the advantage of taking social information into account by conditioning also on the motion of their dancing partner. We focus our analysis on Swing, a dance genre with tight physical coupling for which we present an in-the-wild video dataset. We demonstrate that single-person future motion prediction in this context is challenging. Instead, we observe that prediction greatly benefits from considering the interaction partners' behavior, resulting in surprisingly compelling couple dance synthesis results (see supp. video). Our contributions are a demonstration of the advantages of socially conditioned future motion prediction and an in-the-wild, couple dance video dataset to enable future research in this direction. Video results are available on the project website: https://von31.github.io/synNsync
Authors: Federico Spagnolo, Nataliia Molchanova, Mario Ocampo Pineda, Lester Melie-Garcia, Meritxell Bach Cuadra, Cristina Granziera, Vincent Andrearczyk, Adrien Depeursinge
Abstract: To date, several methods have been developed to explain deep learning algorithms for classification tasks. Recently, an adaptation of two of such methods has been proposed to generate instance-level explainable maps in a semantic segmentation scenario, such as multiple sclerosis (MS) lesion segmentation. In the mentioned work, a 3D U-Net was trained and tested for MS lesion segmentation, yielding an F1 score of 0.7006, and a positive predictive value (PPV) of 0.6265. The distribution of values in explainable maps exposed some differences between maps of true and false positive (TP/FP) examples. Inspired by those results, we explore in this paper the use of characteristics of lesion-specific saliency maps to refine segmentation and detection scores. We generate around 21000 maps from as many TP/FP lesions in a batch of 72 patients (training set) and 4868 from the 37 patients in the test set. 93 radiomic features extracted from the first set of maps were used to train a logistic regression model and classify TP versus FP. On the test set, F1 score and PPV were improved by a large margin when compared to the initial model, reaching 0.7450 and 0.7817, with 95% confidence intervals of [0.7358, 0.7547] and [0.7679, 0.7962], respectively. These results suggest that saliency maps can be used to refine prediction scores, boosting a model's performances.
Authors: Tanish Jain
Abstract: This study evaluates the reliability of two deep learning models for skin cancer detection, focusing on their explainability and fairness. Using the HAM10000 dataset of dermatoscopic images, the research assesses two convolutional neural network architectures: a MobileNet-based model and a custom CNN model. Both models are evaluated for their ability to classify skin lesions into seven categories and to distinguish between dangerous and benign lesions. Explainability is assessed using Saliency Maps and Integrated Gradients, with results interpreted by a dermatologist. The study finds that both models generally highlight relevant features for most lesion types, although they struggle with certain classes like seborrheic keratoses and vascular lesions. Fairness is evaluated using the Equalized Odds metric across sex and skin tone groups. While both models demonstrate fairness across sex groups, they show significant disparities in false positive and false negative rates between light and dark skin tones. A Calibrated Equalized Odds postprocessing strategy is applied to mitigate these disparities, resulting in improved fairness, particularly in reducing false negative rate differences. The study concludes that while the models show promise in explainability, further development is needed to ensure fairness across different skin tones. These findings underscore the importance of rigorous evaluation of AI models in medical applications, particularly in diverse population groups.
Authors: Yudara Kularathne, Prathapa Janitha, Sithira Ambepitiya
Abstract: Background: The 2024 Mpox outbreak, particularly severe in Africa with clade 1b emergence, has highlighted critical gaps in diagnostic capabilities in resource-limited settings. This study aimed to develop and validate an artificial intelligence (AI)-driven, on-device screening tool for Mpox, designed to function offline in low-resource environments. Methods: We developed a YOLOv8n-based deep learning model trained on 2,700 images (900 each of Mpox, other skin conditions, and normal skin), including synthetic data. The model was validated on 360 images and tested on 540 images. A larger external validation was conducted using 1,500 independent images. Performance metrics included accuracy, precision, recall, F1-score, sensitivity, and specificity. Findings: The model demonstrated high accuracy (96%) in the final test set. For Mpox detection, it achieved 93% precision, 97% recall, and an F1-score of 95%. Sensitivity and specificity for Mpox detection were 97% and 96%, respectively. Performance remained consistent in the larger external validation, confirming the model's robustness and generalizability. Interpretation: This AI-driven screening tool offers a rapid, accurate, and scalable solution for Mpox detection in resource-constrained settings. Its offline functionality and high performance across diverse datasets suggest significant potential for improving Mpox surveillance and management, particularly in areas lacking traditional diagnostic infrastructure.
Authors: Karthik Gopinath, Douglas N. Greve, Colin Magdamo, Steve Arnold, Sudeshna Das, Oula Puonti, Juan Eugenio Iglesias
Abstract: Surface-based analysis of the cerebral cortex is ubiquitous in human neuroimaging with MRI. It is crucial for cortical registration, parcellation, and thickness estimation. Traditionally, these analyses require high-resolution, isotropic scans with good gray-white matter contrast, typically a 1mm T1-weighted scan. This excludes most clinical MRI scans, which are often anisotropic and lack the necessary T1 contrast. To enable large-scale neuroimaging studies using vast clinical data, we introduce recon-all-clinical, a novel method for cortical reconstruction, registration, parcellation, and thickness estimation in brain MRI scans of any resolution and contrast. Our approach employs a hybrid analysis method that combines a convolutional neural network (CNN) trained with domain randomization to predict signed distance functions (SDFs) and classical geometry processing for accurate surface placement while maintaining topological and geometric constraints. The method does not require retraining for different acquisitions, thus simplifying the analysis of heterogeneous clinical datasets. We tested recon-all-clinical on multiple datasets, including over 19,000 clinical scans. The method consistently produced precise cortical reconstructions and high parcellation accuracy across varied MRI contrasts and resolutions. Cortical thickness estimates are precise enough to capture aging effects independently of MRI contrast, although accuracy varies with slice thickness. Our method is publicly available at https://surfer.nmr.mgh.harvard.edu/fswiki/recon-all-clinical, enabling researchers to perform detailed cortical analysis on the huge amounts of already existing clinical MRI scans. This advancement may be particularly valuable for studying rare diseases and underrepresented populations where research-grade MRI data is scarce.
URLs: https://surfer.nmr.mgh.harvard.edu/fswiki/recon-all-clinical,
Authors: Zhe Xiong, Qiaoqiao Ding, Xiaoqun Zhang
Abstract: Recently, medical image synthesis gains more and more popularity, along with the rapid development of generative models. Medical image synthesis aims to generate an unacquired image modality, often from other observed data modalities. Synthesized images can be used for clinical diagnostic assistance, data augmentation for model training and validation or image quality improving. In the meanwhile, the flow-based models are among the successful generative models for the ability of generating realistic and high-quality synthetic images. However, most flow-based models require to calculate flow ordinary different equation (ODE) evolution steps in transfer process, for which the performances are significantly limited by heavy computation time due to a large number of time iterations. In this paper, we propose a novel flow-based model, namely Discrete Process Matching (DPM) to accomplish the bi-modality image transfer tasks. Different to other flow matching based models, we propose to utilize both forward and backward ODE flow and enhance the consistency on the intermediate images of few discrete time steps, resulting in a transfer process with much less iteration steps while maintaining high-quality generations for both modalities. Our experiments on three datasets of MRI T1/T2 and CT/MRI demonstrate that DPM outperforms other state-of-the-art flow-based methods for bi-modality image synthesis, achieving higher image quality with less computation time cost.
Authors: Xi Su, Xiangfei Shen, Mingyang Wan, Jing Nie, Lihui Chen, Haijun Liu, Xichuan Zhou
Abstract: Single hyperspectral image super-resolution (single-HSI-SR) aims to improve the resolution of a single input low-resolution HSI. Due to the bottleneck of data scarcity, the development of single-HSI-SR lags far behind that of RGB natural images. In recent years, research on RGB SR has shown that models pre-trained on large-scale benchmark datasets can greatly improve performance on unseen data, which may stand as a remedy for HSI. But how can we transfer the pre-trained RGB model to HSI, to overcome the data-scarcity bottleneck? Because of the significant difference in the channels between the pre-trained RGB model and the HSI, the model cannot focus on the correlation along the spectral dimension, thus limiting its ability to utilize on HSI. Inspired by the HSI spatial-spectral decoupling, we propose a new framework that first fine-tunes the pre-trained model with the spatial components (known as eigenimages), and then infers on unseen HSI using an iterative spectral regularization (ISR) to maintain the spectral correlation. The advantages of our method lie in: 1) we effectively inject the spatial texture processing capabilities of the pre-trained RGB model into HSI while keeping spectral fidelity, 2) learning in the spectral-decorrelated domain can improve the generalizability to spectral-agnostic data, and 3) our inference in the eigenimage domain naturally exploits the spectral low-rank property of HSI, thereby reducing the complexity. This work bridges the gap between pre-trained RGB models and HSI via eigenimages, addressing the issue of limited HSI training data, hence the name EigenSR. Extensive experiments show that EigenSR outperforms the state-of-the-art (SOTA) methods in both spatial and spectral metrics. Our code will be released.
Authors: Phairot Autthasan, Rattanaphon Chaisaen, Huy Phan, Maarten De Vos, Theerawit Wilaiprasitporn
Abstract: Recent advances in deep learning (DL) have significantly impacted motor imagery (MI)-based brain-computer interface (BCI) systems, enhancing the decoding of electroencephalography (EEG) signals. However, most studies struggle to identify discriminative patterns across subjects during MI tasks, limiting MI classification performance. In this article, we propose MixNet, a novel classification framework designed to overcome this limitation by utilizing spectral-spatial signals from MI data, along with a multitask learning architecture named MIN2Net, for classification. Here, the spectral-spatial signals are generated using the filter-bank common spatial patterns (FBCSPs) method on MI data. Since the multitask learning architecture is used for the classification task, the learning in each task may exhibit different generalization rates and potential overfitting across tasks. To address this issue, we implement adaptive gradient blending, simultaneously regulating multiple loss weights and adjusting the learning pace for each task based on its generalization/overfitting tendencies. Experimental results on six benchmark data sets of different data sizes demonstrate that MixNet consistently outperforms all state-of-the-art algorithms in subject-dependent and -independent settings. Finally, the low-density EEG MI classification results show that MixNet outperforms all state-of-the-art algorithms, offering promising implications for Internet of Thing (IoT) applications, such as lightweight and portable EEG wearable devices based on low-density montages.
Authors: Pooja Bidwai, Shilpa Gite, Biswajeet Pradhan, Aditi Gupta, Kishore pahuja
Abstract: This study presents a dataset consisting of 268 retinal images from 179 individuals, including 133 left-eye and 135 right-eye images, collected from Natasha Eye Care and Research Institute in Pune, Maharashtra, India. The images were captured using a nonmydriatic Optical Coherence Tomography Angiography (OCTA) device, specifically the Optovue Avanti Edition machine as per the protocol mentioned in this paper. Two ophthalmologists then annotated the images. This dataset can be used by researchers and doctors to develop automated diagnostic tools for early detection of diabetic retinopathy (DR).
Authors: Valentina Vadori, Jean-Marie Gra\"ic, Antonella Peruffo, Giulia Vadori, Livio Finos, Enrico Grisan
Abstract: Delineating and classifying individual cells in microscopy tissue images is a complex task, yet it is a pivotal endeavor in various medical and biological investigations. We propose a new deep learning framework (CISCA) for automatic cell instance segmentation and classification in histological slices to support detailed morphological and structural analysis or straightforward cell counting in digital pathology workflows and brain cytoarchitecture studies. At the core of CISCA lies a network architecture featuring a lightweight U-Net with three heads in the decoder. The first head classifies pixels into boundaries between neighboring cells, cell bodies, and background, while the second head regresses four distance maps along four directions. The network outputs from the first and second heads are integrated through a tailored post-processing step, which ultimately yields the segmentation of individual cells. A third head enables simultaneous classification of cells into relevant classes, if required. We showcase the effectiveness of our method using four datasets, including CoNIC, PanNuke, and MoNuSeg, which are publicly available H\&E datasets. Additionally, we introduce CytoDArk0, a novel dataset consisting of Nissl-stained images of the cortex, cerebellum, and hippocampus from mammals belonging to the orders Cetartiodactyla and Primates. We evaluate CISCA in comparison to other state-of-the-art methods, demonstrating CISCA's robustness and accuracy in segmenting and classifying cells across diverse tissue types, magnifications, and staining techniques.
Authors: Coby Penso, Jacob Goldberger
Abstract: This study addresses the problem of calibrating network confidence while adapting a model that was originally trained on a source domain to a target domain using unlabeled samples from the target domain. The absence of labels from the target domain makes it impossible to directly calibrate the adapted network on the target domain. To tackle this challenge, we introduce a calibration procedure that relies on estimating the network's accuracy on the target domain. The network accuracy is first computed on the labeled source data and then is modified to represent the actual accuracy of the model on the target domain. The proposed algorithm calibrates the prediction confidence directly in the target domain by minimizing the disparity between the estimated accuracy and the computed confidence. The experimental results show that our method significantly outperforms existing methods, which rely on importance weighting, across several standard datasets.
Authors: Gregory Szumel, Brian Guo, Darui Lu, Rongze Gui, Tingyu Wang, Nicholas Konz, Maciej A. Mazurowski
Abstract: Purpose: Medical images acquired using different scanners and protocols can differ substantially in their appearance. This phenomenon, scanner domain shift, can result in a drop in the performance of deep neural networks which are trained on data acquired by one scanner and tested on another. This significant practical issue is well-acknowledged, however, no systematic study of the issue is available across different modalities and diagnostic tasks. Materials and Methods: In this paper, we present a broad experimental study evaluating the impact of scanner domain shift on convolutional neural network performance for different automated diagnostic tasks. We evaluate this phenomenon in common radiological modalities, including X-ray, CT, and MRI. Results: We find that network performance on data from a different scanner is almost always worse than on same-scanner data, and we quantify the degree of performance drop across different datasets. Notably, we find that this drop is most severe for MRI, moderate for X-ray, and quite small for CT, on average, which we attribute to the standardized nature of CT acquisition systems which is not present in MRI or X-ray. We also study how injecting varying amounts of target domain data into the training set, as well as adding noise to the training data, helps with generalization. Conclusion: Our results provide extensive experimental evidence and quantification of the extent of performance drop caused by scanner domain shift in deep learning across different modalities, with the goal of guiding the future development of robust deep learning models for medical image analysis.
Authors: Davide Clode da Silva, Marina Musse Bernardes, Nathalia Giacomini Ceretta, Gabriel Vaz de Souza, Gabriel Fonseca Silva, Rafael Heitor Bordini, Soraia Raupp Musse
Abstract: Machine learning has significantly advanced healthcare by aiding in disease prevention and treatment identification. However, accessing patient data can be challenging due to privacy concerns and strict regulations. Generating synthetic, realistic data offers a potential solution for overcoming these limitations, and recent studies suggest that fine-tuning foundation models can produce such data effectively. In this study, we explore the potential of foundation models for generating realistic medical images, particularly chest x-rays, and assess how their performance improves with fine-tuning. We propose using a Latent Diffusion Model, starting with a pre-trained foundation model and refining it through various configurations. Additionally, we performed experiments with input from a medical professional to assess the realism of the images produced by each trained model.
Authors: Davide Dalle Pezze, Eugenia Anello, Chiara Masiero, Gian Antonio Susto
Abstract: Anomaly Detection is a relevant problem that arises in numerous real-world applications, especially when dealing with images. However, there has been little research for this task in the Continual Learning setting. In this work, we introduce a novel approach called SCALE (SCALing is Enough) to perform Compressed Replay in a framework for Anomaly Detection in Continual Learning setting. The proposed technique scales and compresses the original images using a Super Resolution model which, to the best of our knowledge, is studied for the first time in the Continual Learning setting. SCALE can achieve a high level of compression while maintaining a high level of image reconstruction quality. In conjunction with other Anomaly Detection approaches, it can achieve optimal results. To validate the proposed approach, we use a real-world dataset of images with pixel-based anomalies, with the scope to provide a reliable benchmark for Anomaly Detection in the context of Continual Learning, serving as a foundation for further advancements in the field.
Authors: Abu Adnan Sadi, Labib Chowdhury, Nusrat Jahan, Mohammad Newaz Sharif Rafi, Radeya Chowdhury, Faisal Ahamed Khan, Nabeel Mohammed
Abstract: With advances in digital technology, the classification of medical images has become a crucial step for image-based clinical decision support systems. Automatic medical image classification represents a pivotal domain where the use of AI holds the potential to create a significant social impact. However, several challenges act as obstacles to the development of practical and effective solutions. One of these challenges is the prevalent class imbalance problem in most medical imaging datasets. As a result, existing AI techniques, particularly deep-learning-based methodologies, often underperform in such scenarios. In this study, we propose a novel framework called Large Margin aware Focal (LMF) loss to mitigate the class imbalance problem in medical imaging. The LMF loss represents a linear combination of two loss functions optimized by two hyperparameters. This framework harnesses the distinct characteristics of both loss functions by enforcing wider margins for minority classes while simultaneously emphasizing challenging samples found in the datasets. We perform rigorous experiments on three neural network architectures and with four medical imaging datasets. We provide empirical evidence that our proposed framework consistently outperforms other baseline methods, showing an improvement of 2%-9% in macro-f1 scores. Through class-wise analysis of f1 scores, we also demonstrate how the proposed framework can significantly improve performance for minority classes. The results of our experiments show that our proposed framework can perform consistently well across different architectures and datasets. Overall, our study demonstrates a simple and effective approach to addressing the class imbalance problem in medical imaging datasets. We hope our work will inspire new research toward a more generalized approach to medical image classification.
Authors: Xinrui Zhan, Xuyang Chang, Daoyu Li, Rong Yan, Yinuo Zhang, Liheng Bian
Abstract: Image sensing often relies on a high-quality machine vision system with a large field of view and high resolution. It requires fine imaging optics, has high computational costs, and requires a large communication bandwidth between image sensors and computing units. In this paper, we propose a novel image-free sensing framework for resource-efficient image classification, where the required number of measurements can be reduced by up to two orders of magnitude. In the proposed framework for single-pixel detection, the optical field for a target is first scattered by an optical diffuser and then two-dimensionally modulated by a spatial light modulator. The optical diffuser simultaneously serves as a compressor and an encryptor for the target information, effectively narrowing the field of view and improving the system's security. The one-dimensional sequence of intensity values, which is measured with time-varying patterns on the spatial light modulator, is then used to extract semantic information based on end-to-end deep learning. The proposed sensing framework is shown to obtain over a 95\% accuracy at sampling rates of 1% and 5% for classification on the MNIST dataset and the recognition of Chinese license plates, respectively, and the framework is up to 24% more efficient than the approach without an optical diffuser. The proposed framework represents a significant breakthrough in high-throughput machine intelligence for scene analysis with low bandwidth, low costs, and strong encryption.
Authors: Niloufar Fakhfour, Mohammad ShahverdiKondori, Sajjad Hashembeiki, Mohammadjavad Norouzi, Hoda Mohammadzade
Abstract: In this paper, we tackle the problem of video alignment, the process of matching the frames of a pair of videos containing similar actions. The main challenge in video alignment is that accurate correspondence should be established despite the differences in the execution processes and appearances between the two videos. We introduce an unsupervised method for alignment that uses global and local features of the frames. In particular, we introduce effective features for each video frame by means of three machine vision tools: person detection, pose estimation, and VGG network. Then the features are processed and combined to construct a multidimensional time series that represent the video. The resulting time series are used to align videos of the same actions using a novel version of dynamic time warping named Diagonalized Dynamic Time Warping(DDTW). The main advantage of our approach is that no training is required, which makes it applicable for any new type of action without any need to collect training samples for it. Additionally, our approach can be used for framewise labeling of action phases in a dataset with only a few labeled videos. For evaluation, we considered video synchronization and phase classification tasks on the Penn action and subset of UCF101 datasets. Also, for an effective evaluation of the video synchronization task, we present a new metric called Enclosed Area Error(EAE). The results show that our method outperforms previous state-of-the-art methods, such as TCC, and other self-supervised and weakly supervised methods.
Authors: Jiaxi Jiang, Paul Streli, Manuel Meier, Christian Holz
Abstract: Full-body egocentric pose estimation from head and hand poses alone has become an active area of research to power articulate avatar representations on headset-based platforms. However, existing methods over-rely on the indoor motion-capture spaces in which datasets were recorded, while simultaneously assuming continuous joint motion capture and uniform body dimensions. We propose EgoPoser to overcome these limitations with four main contributions. 1) EgoPoser robustly models body pose from intermittent hand position and orientation tracking only when inside a headset's field of view. 2) We rethink input representations for headset-based ego-pose estimation and introduce a novel global motion decomposition method that predicts full-body pose independent of global positions. 3) We enhance pose estimation by capturing longer motion time series through an efficient SlowFast module design that maintains computational efficiency. 4) EgoPoser generalizes across various body shapes for different users. We experimentally evaluate our method and show that it outperforms state-of-the-art methods both qualitatively and quantitatively while maintaining a high inference speed of over 600fps. EgoPoser establishes a robust baseline for future work where full-body pose estimation no longer needs to rely on outside-in capture and can scale to large-scale and unseen environments.
Authors: H. Umut Suluhan, Abdullah Enes Doruk, Hasan F. Ates, Bahadir K. Gunturk
Abstract: High-spatio-temporal resolution (HSTR) video recording plays a crucial role in enhancing various imagery tasks that require fine-detailed information. State-of-the-art cameras provide this required high frame-rate and high spatial resolution together, albeit at a high cost. To alleviate this issue, this paper proposes a dual camera system for the generation of HSTR video using reference-based super-resolution (RefSR). One camera captures high spatial resolution low frame rate (HSLF) video while the other captures low spatial resolution high frame rate (LSHF) video simultaneously for the same scene. A novel deep learning architecture is proposed to fuse HSLF and LSHF video feeds and synthesize HSTR video frames. The proposed model combines optical flow estimation and (channel-wise and spatial) attention mechanisms to capture the fine motion and complex dependencies between frames of the two video feeds. Simulations show that the proposed model provides significant improvement over existing reference-based SR techniques in terms of PSNR and SSIM metrics. The method also exhibits sufficient frames per second (FPS) for aerial monitoring when deployed on a power-constrained drone equipped with dual cameras.
Authors: Kyungho Bae, Geo Ahn, Youngrae Kim, Jinwoo Choi
Abstract: Video recognition models often learn scene-biased action representation due to the spurious correlation between actions and scenes in the training data. Such models show poor performance when the test data consists of videos with unseen action-scene combinations. Although scene-debiased action recognition models might address the issue, they often overlook valuable scene information in the data. To address this challenge, we propose to learn DisEntangled VIdeo representations of Action and Scene (DEVIAS), for more holistic video understanding. We propose an encoder-decoder architecture to learn disentangled action and scene representations with a single model. The architecture consists of a disentangling encoder (DE), an action mask decoder (AMD), and a prediction head. The key to achieving the disentanglement is employing both DE and AMD during training time. The DE uses the slot attention mechanism to learn disentangled action and scene representations. For further disentanglement, an AMD learns to predict action masks, given an action slot. With the resulting disentangled representations, we can achieve robust performance across diverse scenarios, including both seen and unseen action-scene combinations. We rigorously validate the proposed method on the UCF-101, Kinetics-400, and HVU datasets for the seen, and the SCUBA, HAT, and HVU datasets for unseen action-scene combination scenarios. Furthermore, DEVIAS provides flexibility to adjust the emphasis on action or scene information depending on dataset characteristics for downstream tasks. DEVIAS shows favorable performance in various downstream tasks: Diving48, Something-Something-V2, UCF-101, and ActivityNet. The code is available at https://github.com/KHU-VLL/DEVIAS.
Authors: Can Jin, Tianjin Huang, Yihua Zhang, Mykola Pechenizkiy, Sijia Liu, Shiwei Liu, Tianlong Chen
Abstract: The rapid development of large-scale deep learning models questions the affordability of hardware platforms, which necessitates the pruning to reduce their computational and memory footprints. Sparse neural networks as the product, have demonstrated numerous favorable benefits like low complexity, undamaged generalization, etc. Most of the prominent pruning strategies are invented from a model-centric perspective, focusing on searching and preserving crucial weights by analyzing network topologies. However, the role of data and its interplay with model-centric pruning has remained relatively unexplored. In this research, we introduce a novel data-model co-design perspective: to promote superior weight sparsity by learning important model topology and adequate input data in a synergetic manner. Specifically, customized Visual Prompts are mounted to upgrade neural Network sparsification in our proposed VPNs framework. As a pioneering effort, this paper conducts systematic investigations about the impact of different visual prompts on model pruning and suggests an effective joint optimization approach. Extensive experiments with 3 network architectures and 8 datasets evidence the substantial performance improvements from VPNs over existing start-of-the-art pruning algorithms. Furthermore, we find that subnetworks discovered by VPNs from pre-trained models enjoy better transferability across diverse downstream scenarios. These insights shed light on new promising possibilities of data-model co-designs for vision model sparsification.
Authors: Ryan Po, Guandao Yang, Kfir Aberman, Gordon Wetzstein
Abstract: Customization techniques for text-to-image models have paved the way for a wide range of previously unattainable applications, enabling the generation of specific concepts across diverse contexts and styles. While existing methods facilitate high-fidelity customization for individual concepts or a limited, pre-defined set of them, they fall short of achieving scalability, where a single model can seamlessly render countless concepts. In this paper, we address a new problem called Modular Customization, with the goal of efficiently merging customized models that were fine-tuned independently for individual concepts. This allows the merged model to jointly synthesize concepts in one image without compromising fidelity or incurring any additional computational costs. To address this problem, we introduce Orthogonal Adaptation, a method designed to encourage the customized models, which do not have access to each other during fine-tuning, to have orthogonal residual weights. This ensures that during inference time, the customized models can be summed with minimal interference. Our proposed method is both simple and versatile, applicable to nearly all optimizable weights in the model architecture. Through an extensive set of quantitative and qualitative evaluations, our method consistently outperforms relevant baselines in terms of efficiency and identity preservation, demonstrating a significant leap toward scalable customization of diffusion models.
Authors: Pau Torras, Sanket Biswas, Alicia Forn\'es
Abstract: Modern-day Optical Music Recognition (OMR) is a fairly fragmented field. Most OMR approaches use datasets that are independent and incompatible between each other, making it difficult to both combine them and compare recognition systems built upon them. In this paper we identify the need of a common music representation language and propose the Music Tree Notation (MTN) format, with the idea to construct a common endpoint for OMR research that allows coordination, reuse of technology and fair evaluation of community efforts. This format represents music as a set of primitives that group together into higher-abstraction nodes, a compromise between the expression of fully graph-based and sequential notation formats. We have also developed a specific set of OMR metrics and a typeset score dataset as a proof of concept of this idea.
Authors: Xiaoliang Tan, Guanzhou Chen, Tong Wang, Jiaqi Wang, Xiaodong Zhang
Abstract: The field of Remote Sensing (RS) widely employs Change Detection (CD) on very-high-resolution (VHR) images. A majority of extant deep-learning-based methods hinge on annotated samples to complete the CD process. Recently, the emergence of Vision Foundation Model (VFM) enables zero-shot predictions in particular vision tasks. In this work, we propose an unsupervised CD method named Segment Change Model (SCM), built upon the Segment Anything Model (SAM) and Contrastive Language-Image Pre-training (CLIP). Our method recalibrates features extracted at different scales and integrates them in a top-down manner to enhance discriminative change edges. We further design an innovative Piecewise Semantic Attention (PSA) scheme, which can offer semantic representation without training, thereby minimize pseudo change phenomenon. Through conducting experiments on two public datasets, the proposed SCM increases the mIoU from 46.09% to 53.67% on the LEVIR-CD dataset, and from 47.56% to 52.14% on the WHU-CD dataset. Our codes are available at https://github.com/StephenApX/UCD-SCM.
Authors: Haoxuan Wang, Yuzhang Shang, Zhihang Yuan, Junyi Wu, Junchi Yan, Yan Yan
Abstract: The practical deployment of diffusion models still suffers from the high memory and time overhead. While quantization paves a way for compression and acceleration, existing methods unfortunately fail when the models are quantized to low-bits. In this paper, we empirically unravel three properties in quantized diffusion models that compromise the efficacy of current methods: imbalanced activation distributions, imprecise temporal information, and vulnerability to perturbations of specific modules. To alleviate the intensified low-bit quantization difficulty stemming from the distribution imbalance, we propose finetuning the quantized model to better adapt to the activation distribution. Building on this idea, we identify two critical types of quantized layers: those holding vital temporal information and those sensitive to reduced bit-width, and finetune them to mitigate performance degradation with efficiency. We empirically verify that our approach modifies the activation distribution and provides meaningful temporal information, facilitating easier and more accurate quantization. Our method is evaluated over three high-resolution image generation tasks and achieves state-of-the-art performance under various bit-width settings, as well as being the first method to generate readable images on full 4-bit (i.e. W4A4) Stable Diffusion. Code is available \href{https://github.com/hatchetProject/QuEST}{here}.
Authors: Ruoyu Chen, Hua Zhang, Siyuan Liang, Jingzhi Li, Xiaochun Cao
Abstract: Image attribution algorithms aim to identify important regions that are highly relevant to model decisions. Although existing attribution solutions can effectively assign importance to target elements, they still face the following challenges: 1) existing attribution methods generate inaccurate small regions thus misleading the direction of correct attribution, and 2) the model cannot produce good attribution results for samples with wrong predictions. To address the above challenges, this paper re-models the above image attribution problem as a submodular subset selection problem, aiming to enhance model interpretability using fewer regions. To address the lack of attention to local regions, we construct a novel submodular function to discover more accurate small interpretation regions. To enhance the attribution effect for all samples, we also impose four different constraints on the selection of sub-regions, i.e., confidence, effectiveness, consistency, and collaboration scores, to assess the importance of various subsets. Moreover, our theoretical analysis substantiates that the proposed function is in fact submodular. Extensive experiments show that the proposed method outperforms SOTA methods on two face datasets (Celeb-A and VGG-Face2) and one fine-grained dataset (CUB-200-2011). For correctly predicted samples, the proposed method improves the Deletion and Insertion scores with an average of 4.9% and 2.5% gain relative to HSIC-Attribution. For incorrectly predicted samples, our method achieves gains of 81.0% and 18.4% compared to the HSIC-Attribution algorithm in the average highest confidence and Insertion score respectively. The code is released at https://github.com/RuoyuChen10/SMDL-Attribution.
Authors: Daniel DeAlcala, Aythami Morales, Julian Fierrez, Gonzalo Mancera, Ruben Tolosana, Javier Ortega-Garcia
Abstract: This article introduces the Membership Inference Test (MINT), a novel approach that aims to empirically assess if given data was used during the training of AI/ML models. Specifically, we propose two MINT architectures designed to learn the distinct activation patterns that emerge when an Audited Model is exposed to data used during its training process. These architectures are based on Multilayer Perceptrons (MLPs) and Convolutional Neural Networks (CNNs). The experimental framework focuses on the challenging task of Face Recognition, considering three state-of-the-art Face Recognition systems. Experiments are carried out using six publicly available databases, comprising over 22 million face images in total. Different experimental scenarios are considered depending on the context of the AI model to test. Our proposed MINT approach achieves promising results, with up to 90% accuracy, indicating the potential to recognize if an AI model has been trained with specific data. The proposed MINT approach can serve to enforce privacy and fairness in several AI applications, e.g., revealing if sensitive or private data was used for training or tuning Large Language Models (LLMs).
Authors: Chi-Sheng Chen, Guan-Ying Chen, Dong Zhou, Di Jiang, Dai-Shi Chen
Abstract: Food classification is the foundation for developing food vision tasks and plays a key role in the burgeoning field of computational nutrition. Due to the complexity of food requiring fine-grained classification, recent academic research mainly modifies Convolutional Neural Networks (CNNs) and/or Vision Transformers (ViTs) to perform food category classification. However, to learn fine-grained features, the CNN backbone needs additional structural design, whereas ViT, containing the self-attention module, has increased computational complexity. In recent months, a new Sequence State Space (S4) model, through a Selection mechanism and computation with a Scan (S6), colloquially termed Mamba, has demonstrated superior performance and computation efficiency compared to the Transformer architecture. The VMamba model, which incorporates the Mamba mechanism into image tasks (such as classification), currently establishes the state-of-the-art (SOTA) on the ImageNet dataset. In this research, we introduce an academically underestimated food dataset CNFOOD-241, and pioneer the integration of a residual learning framework within the VMamba model to concurrently harness both global and local state features inherent in the original VMamba architectural design. The research results show that VMamba surpasses current SOTA models in fine-grained and food classification. The proposed Res-VMamba further improves the classification accuracy to 79.54\% without pretrained weight. Our findings elucidate that our proposed methodology establishes a new benchmark for SOTA performance in food recognition on the CNFOOD-241 dataset. The code can be obtained on GitHub: https://github.com/ChiShengChen/ResVMamba.
Authors: Hanning Chen, Wenjun Huang, Yang Ni, Sanggeon Yun, Yezi Liu, Fei Wen, Alvaro Velasquez, Hugo Latapie, Mohsen Imani
Abstract: Task-oriented object detection aims to find objects suitable for accomplishing specific tasks. As a challenging task, it requires simultaneous visual data processing and reasoning under ambiguous semantics. Recent solutions are mainly all-in-one models. However, the object detection backbones are pre-trained without text supervision. Thus, to incorporate task requirements, their intricate models undergo extensive learning on a highly imbalanced and scarce dataset, resulting in capped performance, laborious training, and poor generalizability. In contrast, we propose TaskCLIP, a more natural two-stage design composed of general object detection and task-guided object selection. Particularly for the latter, we resort to the recently successful large Vision-Language Models (VLMs) as our backbone, which provides rich semantic knowledge and a uniform embedding space for images and texts. Nevertheless, the naive application of VLMs leads to sub-optimal quality, due to the misalignment between embeddings of object images and their visual attributes, which are mainly adjective phrases. To this end, we design a transformer-based aligner after the pre-trained VLMs to re-calibrate both embeddings. Finally, we employ a trainable score function to post-process the VLM matching results for object selection. Experimental results demonstrate that our TaskCLIP outperforms the state-of-the-art DETR-based model TOIST by 3.5% and only requires a single NVIDIA RTX 4090 for both training and inference.
Authors: Abhishek Saroha, Mariia Gladkova, Cecilia Curreli, Dominik Muhle, Tarun Yenamandra, Daniel Cremers
Abstract: 3D scene stylization extends the work of neural style transfer to 3D. A vital challenge in this problem is to maintain the uniformity of the stylized appearance across multiple views. A vast majority of the previous works achieve this by training a 3D model for every stylized image and a set of multi-view images. In contrast, we propose a novel architecture trained on a collection of style images that, at test time, produces real time high-quality stylized novel views. We choose the underlying 3D scene representation for our model as 3D Gaussian splatting. We take the 3D Gaussians and process them using a multi-resolution hash grid and a tiny MLP to obtain stylized views. The MLP is conditioned on different style codes for generalization to different styles during test time. The explicit nature of 3D Gaussians gives us inherent advantages over NeRF-based methods, including geometric consistency and a fast training and rendering regime. This enables our method to be useful for various practical use cases, such as augmented or virtual reality. We demonstrate that our method achieves state-of-the-art performance with superior visual quality on various indoor and outdoor real-world data.
Authors: Zhixiu Lu, Hailong Li, Nehal A. Parikh, Jonathan R. Dillman, Lili He
Abstract: The integration of artificial intelligence (AI) with radiology marks a transformative era in medicine. Vision foundation models have been adopted to enhance radiologic imaging analysis. However, the distinct complexities of radiologic 2D and 3D radiologic data pose unique challenges that existing models, pre-trained on general non-medical images, fail to address adequately. To bridge this gap and capitalize on the diagnostic precision required in radiologic imaging, we introduce Radiologic Contrastive Language-Image Pre-training (RadCLIP): a cross-modal vision-language foundational model that harnesses Vision Language Pre-training (VLP) framework to improve radiologic image analysis. Building upon Contrastive Language-Image Pre-training (CLIP), RadCLIP incorporates a slice pooling mechanism tailored for volumetric image analysis and is pre-trained using a large and diverse dataset of radiologic image-text pairs. The RadCLIP was pre-trained to effectively align radiologic images with their corresponding text annotations, creating a robust vision backbone for radiologic images. Extensive experiments demonstrate RadCLIP's superior performance in both uni-modal radiologic image classification and cross-modal image-text matching, highlighting its significant promise for improving diagnostic accuracy and efficiency in clinical settings. Our Key contributions include curating a large dataset with diverse radiologic 2D/3D radiologic image-text pairs, a slice pooling adapter using an attention mechanism for integrating 2D images, and comprehensive evaluations of RadCLIP on various radiologic downstream tasks.
Authors: Jue Wang, Yuxiang Lin, Qi Zhao, Dong Luo, Shuaibao Chen, Wei Chen, Xiaojiang Peng
Abstract: The widespread use of various chemical gases in industrial processes necessitates effective measures to prevent their leakage during transportation and storage, given their high toxicity. Thermal infrared-based computer vision detection techniques provide a straightforward approach to identify gas leakage areas. However, the development of high-quality algorithms has been challenging due to the low texture in thermal images and the lack of open-source datasets. In this paper, we present the RGB-Thermal Cross Attention Network (RT-CAN), which employs an RGB-assisted two-stream network architecture to integrate texture information from RGB images and gas area information from thermal images. Additionally, to facilitate the research of invisible gas detection, we introduce Gas-DB, an extensive open-source gas detection database including about 1.3K well-annotated RGB-thermal images with eight variant collection scenes. Experimental results demonstrate that our method successfully leverages the advantages of both modalities, achieving state-of-the-art (SOTA) performance among RGB-thermal methods, surpassing single-stream SOTA models in terms of accuracy, Intersection of Union (IoU), and F2 metrics by 4.86%, 5.65%, and 4.88%, respectively. The code and data can be found at https://github.com/logic112358/RT-CAN.
Authors: Zishu Yao, Guodong Fan, Jinfu Fan, Min Gan, C. L. Philip Chen
Abstract: Low-light remote sensing images generally feature high resolution and high spatial complexity, with continuously distributed surface features in space. This continuity in scenes leads to extensive long-range correlations in spatial domains within remote sensing images. Convolutional Neural Networks, which rely on local correlations for long-distance modeling, struggle to establish long-range correlations in such images. On the other hand, transformer-based methods that focus on global information face high computational complexities when processing high-resolution remote sensing images. From another perspective, Fourier transform can compute global information without introducing a large number of parameters, enabling the network to more efficiently capture the overall image structure and establish long-range correlations. Therefore, we propose a Dual-Domain Feature Fusion Network (DFFN) for low-light remote sensing image enhancement. Specifically, this challenging task of low-light enhancement is divided into two more manageable sub-tasks: the first phase learns amplitude information to restore image brightness, and the second phase learns phase information to refine details. To facilitate information exchange between the two phases, we designed an information fusion affine block that combines data from different phases and scales. Additionally, we have constructed two dark light remote sensing datasets to address the current lack of datasets in dark light remote sensing image enhancement. Extensive evaluations show that our method outperforms existing state-of-the-art methods. The code is available at https://github.com/iijjlk/DFFN.
Authors: Zonglin Lyu, Ming Li, Jianbo Jiao, Chen Chen
Abstract: Recent work in Video Frame Interpolation (VFI) tries to formulate VFI as a diffusion-based conditional image generation problem, synthesizing the intermediate frame given a random noise and neighboring frames. Due to the relatively high resolution of videos, Latent Diffusion Models (LDMs) are employed as the conditional generation model, where the autoencoder compresses images into latent representations for diffusion and then reconstructs images from these latent representations. Such a formulation poses a crucial challenge: VFI expects that the output is deterministically equal to the ground truth intermediate frame, but LDMs randomly generate a diverse set of different images when the model runs multiple times. The reason for the diverse generation is that the cumulative variance (variance accumulated at each step of generation) of generated latent representations in LDMs is large. This makes the sampling trajectory random, resulting in diverse rather than deterministic generations. To address this problem, we propose our unique solution: Frame Interpolation with Consecutive Brownian Bridge Diffusion. Specifically, we propose consecutive Brownian Bridge diffusion that takes a deterministic initial value as input, resulting in a much smaller cumulative variance of generated latent representations. Our experiments suggest that our method can improve together with the improvement of the autoencoder and achieve state-of-the-art performance in VFI, leaving strong potential for further enhancement.
Authors: Prakash Chandra Chhipa, Kanjar De, Meenakshi Subhash Chippa, Rajkumar Saini, Marcus Liwicki
Abstract: The challenge of Out-Of-Distribution (OOD) robustness remains a critical hurdle towards deploying deep vision models. Vision-Language Models (VLMs) have recently achieved groundbreaking results. VLM-based open-vocabulary object detection extends the capabilities of traditional object detection frameworks, enabling the recognition and classification of objects beyond predefined categories. Investigating OOD robustness in recent open-vocabulary object detection is essential to increase the trustworthiness of these models. This study presents a comprehensive robustness evaluation of the zero-shot capabilities of three recent open-vocabulary (OV) foundation object detection models: OWL-ViT, YOLO World, and Grounding DINO. Experiments carried out on the robustness benchmarks COCO-O, COCO-DC, and COCO-C encompassing distribution shifts due to information loss, corruption, adversarial attacks, and geometrical deformation, highlighting the challenges of the model's robustness to foster the research for achieving robustness. Project page: https://prakashchhipa.github.io/projects/ovod_robustness
URLs: https://prakashchhipa.github.io/projects/ovod_robustness
Authors: Lianghan Zhu, Yanqi Bao, Jing Huo, Jing Wu, Yu-Kun Lai, Wenbin Li, Yang Gao
Abstract: The rapidly evolving field of Text-to-Video generation (T2V) has catalyzed renewed interest in controllable video editing research. While the application of editing prompts to guide diffusion model denoising has gained prominence, mirroring advancements in image editing, this noise-based inference process inherently compromises the original video's integrity, resulting in unintended over-editing and temporal discontinuities. To address these challenges, this study proposes a novel paradigm of video-based score distillation, facilitating direct manipulation of original video content. Specifically, distinguishing it from image-based score distillation, we propose an Adaptive Sliding Score Distillation strategy, which incorporates both global and local video guidance to reduce the impact of editing errors. Combined with our proposed Image-based Joint Guidance mechanism, it has the ability to mitigate the inherent instability of the T2V model and single-step sampling. Additionally, we design a Weighted Attention Fusion module to further preserve the key features of the original video and avoid over-editing. Extensive experiments demonstrate that these strategies effectively address existing challenges, achieving superior performance compared to current state-of-the-art methods.
Authors: Chen Hou, Guoqiang Wei, Yan Zeng, Zhibo Chen
Abstract: We propose a training-free and robust solution to offer camera movement control for off-the-shelf video diffusion models. Unlike previous work, our method does not require any supervised finetuning on camera-annotated datasets or self-supervised training via data augmentation. Instead, it can be plugged and played with most pretrained video diffusion models and generate camera controllable videos with a single image or text prompt as input. The inspiration of our work comes from the layout prior that intermediate latents hold towards generated results, thus rearranging noisy pixels in them will make output content reallocated as well. As camera move could also be seen as a kind of pixel rearrangement caused by perspective change, videos could be reorganized following specific camera motion if their noisy latents change accordingly. Established on this, we propose our method CamTrol, which enables robust camera control for video diffusion models. It is achieved by a two-stage process. First, we model image layout rearrangement through explicit camera movement in 3D point cloud space. Second, we generate videos with camera motion using layout prior of noisy latents formed by a series of rearranged images. Extensive experiments have demonstrated the robustness our method holds in controlling camera motion of generated videos. Furthermore, we show that our method can produce impressive results in generating 3D rotation videos with dynamic content. Project page at https://lifedecoder.github.io/CamTrol/.
Authors: Valeriy Lobanov, Nikita Firsov, Evgeny Myasnikov, Roman Khabibullin, Artem Nikonorov
Abstract: In traditional neural network architectures, a multilayer perceptron (MLP) is typically employed as a classification block following the feature extraction stage. However, the Kolmogorov-Arnold Network (KAN) presents a promising alternative to MLP, offering the potential to enhance prediction accuracy. In this paper, we propose the replacement of linear and convolutional layers of traditional networks with KAN-based counterparts. These modifications allowed us to significantly increase the per-pixel classification accuracy for hyperspectral remote-sensing images. We modified seven different neural network architectures for hyperspectral image classification and observed a substantial improvement in the classification accuracy across all the networks. The architectures considered in the paper include baseline MLP, state-of-the-art 1D (1DCNN) and 3D convolutional (two different 3DCNN, NM3DCNN), and transformer (SSFTT) architectures, as well as newly proposed M1DCNN. The greatest effect was achieved for convolutional networks working exclusively on spectral data, and the best classification quality was achieved using a KAN-based transformer architecture. All the experiments were conducted using seven openly available hyperspectral datasets. Our code is available at https://github.com/f-neumann77/HyperKAN.
Authors: Ningli Xu, Rongjun Qin
Abstract: Predicting realistic ground views from satellite imagery in urban scenes is a challenging task due to the significant view gaps between satellite and ground-view images. We propose a novel pipeline to tackle this challenge, by generating geospecifc views that maximally respect the weak geometry and texture from multi-view satellite images. Different from existing approaches that hallucinate images from cues such as partial semantics or geometry from overhead satellite images, our method directly predicts ground-view images at geolocation by using a comprehensive set of information from the satellite image, resulting in ground-level images with a resolution boost at a factor of ten or more. We leverage a novel building refinement method to reduce geometric distortions in satellite data at ground level, which ensures the creation of accurate conditions for view synthesis using diffusion networks. Moreover, we proposed a novel geospecific prior, which prompts distribution learning of diffusion models to respect image samples that are closer to the geolocation of the predicted images. We demonstrate our pipeline is the first to generate close-to-real and geospecific ground views merely based on satellite images.
Authors: Daniel Braun, Remco Chang, Michael Gleicher, Tatiana von Landesberger
Abstract: Visual validation of regression models in scatterplots is a common practice for assessing model quality, yet its efficacy remains unquantified. We conducted two empirical experiments to investigate individuals' ability to visually validate linear regression models (linear trends) and to examine the impact of common visualization designs on validation quality. The first experiment showed that the level of accuracy for visual estimation of slope (i.e., fitting a line to data) is higher than for visual validation of slope (i.e., accepting a shown line). Notably, we found bias toward slopes that are "too steep" in both cases. This lead to novel insights that participants naturally assessed regression with orthogonal distances between the points and the line (i.e., ODR regression) rather than the common vertical distances (OLS regression). In the second experiment, we investigated whether incorporating common designs for regression visualization (error lines, bounding boxes, and confidence intervals) would improve visual validation. Even though error lines reduced validation bias, results failed to show the desired improvements in accuracy for any design. Overall, our findings suggest caution in using visual model validation for linear trends in scatterplots.
Authors: Zhengcong Fei, Mingyuan Fan, Changqian Yu, Debang Li, Junshi Huang
Abstract: In this paper, we present DiT-MoE, a sparse version of the diffusion Transformer, that is scalable and competitive with dense networks while exhibiting highly optimized inference. The DiT-MoE includes two simple designs: shared expert routing and expert-level balance loss, thereby capturing common knowledge and reducing redundancy among the different routed experts. When applied to conditional image generation, a deep analysis of experts specialization gains some interesting observations: (i) Expert selection shows preference with spatial position and denoising time step, while insensitive with different class-conditional information; (ii) As the MoE layers go deeper, the selection of experts gradually shifts from specific spacial position to dispersion and balance. (iii) Expert specialization tends to be more concentrated at the early time step and then gradually uniform after half. We attribute it to the diffusion process that first models the low-frequency spatial information and then high-frequency complex information. Based on the above guidance, a series of DiT-MoE experimentally achieves performance on par with dense networks yet requires much less computational load during inference. More encouragingly, we demonstrate the potential of DiT-MoE with synthesized image data, scaling diffusion model at a 16.5B parameter that attains a new SoTA FID-50K score of 1.80 in 512$\times$512 resolution settings. The project page: https://github.com/feizc/DiT-MoE.
Authors: Qiao Wu, Kun Sun, Pei An, Mathieu Salzmann, Yanning Zhang, Jiaqi Yang
Abstract: The high temporal variation of the point clouds is the key challenge of 3D single-object tracking (3D SOT). Existing approaches rely on the assumption that the shape variation of the point clouds and the motion of the objects across neighboring frames are smooth, failing to cope with high temporal variation data. In this paper, we present a novel framework for 3D SOT in point clouds with high temporal variation, called HVTrack. HVTrack proposes three novel components to tackle the challenges in the high temporal variation scenario: 1) A Relative-Pose-Aware Memory module to handle temporal point cloud shape variations; 2) a Base-Expansion Feature Cross-Attention module to deal with similar object distractions in expanded search areas; 3) a Contextual Point Guided Self-Attention module for suppressing heavy background noise. We construct a dataset with high temporal variation (KITTI-HV) by setting different frame intervals for sampling in the KITTI dataset. On the KITTI-HV with 5 frame intervals, our HVTrack surpasses the state-of-the-art tracker CXTracker by 11.3%/15.7% in Success/Precision.
Authors: Junjie He, Yifeng Geng, Liefeng Bo
Abstract: This paper presents UniPortrait, an innovative human image personalization framework that unifies single- and multi-ID customization with high face fidelity, extensive facial editability, free-form input description, and diverse layout generation. UniPortrait consists of only two plug-and-play modules: an ID embedding module and an ID routing module. The ID embedding module extracts versatile editable facial features with a decoupling strategy for each ID and embeds them into the context space of diffusion models. The ID routing module then combines and distributes these embeddings adaptively to their respective regions within the synthesized image, achieving the customization of single and multiple IDs. With a carefully designed two-stage training scheme, UniPortrait achieves superior performance in both single- and multi-ID customization. Quantitative and qualitative experiments demonstrate the advantages of our method over existing approaches as well as its good scalability, e.g., the universal compatibility with existing generative control tools. The project page is at https://aigcdesigngroup.github.io/UniPortrait-Page/ .
Authors: Xiaokun Sun, Zhenyu Zhang, Ying Tai, Qian Wang, Hao Tang, Zili Yi, Jian Yang
Abstract: Recent advances in text-guided 3D avatar generation have made substantial progress by distilling knowledge from diffusion models. Despite the plausible generated appearance, existing methods cannot achieve fine-grained disentanglement or high-fidelity modeling between inner body and outfit. In this paper, we propose Barbie, a novel framework for generating 3D avatars that can be dressed in diverse and high-quality Barbie-like garments and accessories. Instead of relying on a holistic model, Barbie achieves fine-grained disentanglement on avatars by semantic-aligned separated models for human body and outfits. These disentangled 3D representations are then optimized by different expert models to guarantee the domain-specific fidelity. To balance geometry diversity and reasonableness, we propose a series of losses for template-preserving and human-prior evolving. The final avatar is enhanced by unified texture refinement for superior texture consistency. Extensive experiments demonstrate that Barbie outperforms existing methods in both dressed human and outfit generation, supporting flexible apparel combination and animation. The code will be released for research purposes. Our project page is: https://xiaokunsun.github.io/Barbie.github.io/.
Authors: Corentin Dumery, Aoxiang Fan, Ren Li, Nicolas Talabot, Pascal Fua
Abstract: Neural Radiance Fields (NeRFs) have become a powerful tool for modeling 3D scenes from multiple images. However, NeRFs remain difficult to segment into semantically meaningful regions. Previous approaches to 3D segmentation of NeRFs either require user interaction to isolate a single object, or they rely on 2D semantic masks with a limited number of classes for supervision. As a consequence, they generalize poorly to class-agnostic masks automatically generated in real scenes. This is attributable to the ambiguity arising from zero-shot segmentation, yielding inconsistent masks across views. In contrast, we propose a method that is robust to inconsistent segmentations and successfully decomposes the scene into a set of objects of any class. By introducing a limited number of competing object slots against which masks are matched, a meaningful object representation emerges that best explains the 2D supervision and minimizes an additional regularization term. Our experiments demonstrate the ability of our method to generate 3D panoptic segmentations on complex scenes, and extract high-quality 3D assets from NeRFs that can then be used in virtual 3D environments.
Authors: Ziyi Zhang, Nicolas Roussel, Wenzel Jakob
Abstract: Discontinuous visibility changes remain a major bottleneck when optimizing surfaces within a physically-based inverse renderer. Many previous works have proposed sophisticated algorithms and data structures to sample visibility silhouettes more efficiently. Our work presents another solution: instead of differentiating a tentative surface locally, we differentiate a volumetric perturbation of a surface. We refer this as a many-worlds representation because it models a non-interacting superposition of conflicting explanations (worlds) of the input dataset. Each world is optically isolated from others, leading to a new transport law that distinguishes our method from prior work based on exponential random media. The resulting Monte Carlo algorithm is simpler and more efficient than prior methods. We demonstrate that our method promotes rapid convergence, both in terms of the total iteration count and the cost per iteration.
Authors: Nasim Jamshidi Avanaki, Abhijay Ghildyal, Nabajeet Barman, Saman Zadtootaghaj
Abstract: No-Reference Image Quality Assessment (NR-IQA) remains a challenging task due to the diversity of distortions and the lack of large annotated datasets. Many studies have attempted to tackle these challenges by developing more accurate NR-IQA models, often employing complex and computationally expensive networks, or by bridging the domain gap between various distortions to enhance performance on test datasets. In our work, we improve the performance of a generic lightweight NR-IQA model by introducing a novel augmentation strategy that boosts its performance by almost 28\%. This augmentation strategy enables the network to better discriminate between different distortions in various parts of the image by zooming in and out. Additionally, the inclusion of test-time augmentation further enhances performance, making our lightweight network's results comparable to the current state-of-the-art models, simply through the use of augmentations.
Authors: Nasim Jamshidi Avanaki, Abhijay Ghildyal, Nabajeet Barman, Saman Zadtootaghaj
Abstract: Recent advancements in the field of No-Reference Image Quality Assessment (NR-IQA) using deep learning techniques demonstrate high performance across multiple open-source datasets. However, such models are typically very large and complex making them not so suitable for real-world deployment, especially on resource- and battery-constrained mobile devices. To address this limitation, we propose a compact, lightweight NR-IQA model that achieves state-of-the-art (SOTA) performance on ECCV AIM UHD-IQA challenge validation and test datasets while being also nearly 5.7 times faster than the fastest SOTA model. Our model features a dual-branch architecture, with each branch separately trained on synthetically and authentically distorted images which enhances the model's generalizability across different distortion types. To improve robustness under diverse real-world visual conditions, we additionally incorporate multiple color spaces during the training process. We also demonstrate the higher accuracy of recently proposed Kolmogorov-Arnold Networks (KANs) for final quality regression as compared to the conventional Multi-Layer Perceptrons (MLPs). Our evaluation considering various open-source datasets highlights the practical, high-accuracy, and robust performance of our proposed lightweight model. Code: https://github.com/nasimjamshidi/LAR-IQA.
Authors: Zanlin Ni, Yulin Wang, Renping Zhou, Rui Lu, Jiayi Guo, Jinyi Hu, Zhiyuan Liu, Yuan Yao, Gao Huang
Abstract: Recent studies have demonstrated the effectiveness of token-based methods for visual content generation. As a representative work, non-autoregressive Transformers (NATs) are able to synthesize images with decent quality in a small number of steps. However, NATs usually necessitate configuring a complicated generation policy comprising multiple manually-designed scheduling rules. These heuristic-driven rules are prone to sub-optimality and come with the requirements of expert knowledge and labor-intensive efforts. Moreover, their one-size-fits-all nature cannot flexibly adapt to the diverse characteristics of each individual sample. To address these issues, we propose AdaNAT, a learnable approach that automatically configures a suitable policy tailored for every sample to be generated. In specific, we formulate the determination of generation policies as a Markov decision process. Under this framework, a lightweight policy network for generation can be learned via reinforcement learning. Importantly, we demonstrate that simple reward designs such as FID or pre-trained reward models, may not reliably guarantee the desired quality or diversity of generated samples. Therefore, we propose an adversarial reward design to guide the training of policy networks effectively. Comprehensive experiments on four benchmark datasets, i.e., ImageNet-256 & 512, MS-COCO, and CC3M, validate the effectiveness of AdaNAT. Code and pre-trained models will be released at https://github.com/LeapLabTHU/AdaNAT.
Authors: YuanZheng Wu, Jin Liu, Shunping Ji
Abstract: Recently, 3D Gaussian Splatting (3DGS) has garnered significant attention. However, the unstructured nature of 3DGS poses challenges for large-scale surface reconstruction from aerial images. To address this gap, we propose the first large-scale surface reconstruction method for multi-view stereo (MVS) aerial images based on 3DGS, named Aerial Gaussian Splatting (AGS). Initially, we introduce a data chunking method tailored for large-scale aerial imagery, making the modern 3DGS technology feasible for surface reconstruction over extensive scenes. Additionally, we integrate the Ray-Gaussian Intersection method to obtain normal and depth information, facilitating geometric constraints. Finally, we introduce a multi-view geometric consistency constraint to enhance global geometric consistency and improve reconstruction accuracy. Our experiments on multiple datasets demonstrate for the first time that the GS-based technique can match traditional aerial MVS methods on geometric accuracy, and beat state-of-the-art GS-based methods on geometry and rendering quality.
Authors: Xinyu Liu, Yingqing He, Lanqing Guo, Xiang Li, Bu Jin, Peng Li, Yan Li, Chi-Min Chan, Qifeng Chen, Wei Xue, Wenhan Luo, Qingfeng Liu, Yike Guo
Abstract: The potential for higher-resolution image generation using pretrained diffusion models is immense, yet these models often struggle with issues of object repetition and structural artifacts especially when scaling to 4K resolution and higher. We figure out that the problem is caused by that, a single prompt for the generation of multiple scales provides insufficient efficacy. In response, we propose HiPrompt, a new tuning-free solution that tackles the above problems by introducing hierarchical prompts. The hierarchical prompts offer both global and local guidance. Specifically, the global guidance comes from the user input that describes the overall content, while the local guidance utilizes patch-wise descriptions from MLLMs to elaborately guide the regional structure and texture generation. Furthermore, during the inverse denoising process, the generated noise is decomposed into low- and high-frequency spatial components. These components are conditioned on multiple prompt levels, including detailed patch-wise descriptions and broader image-level prompts, facilitating prompt-guided denoising under hierarchical semantic guidance. It further allows the generation to focus more on local spatial regions and ensures the generated images maintain coherent local and global semantics, structures, and textures with high definition. Extensive experiments demonstrate that HiPrompt outperforms state-of-the-art works in higher-resolution image generation, significantly reducing object repetition and enhancing structural quality.
Authors: Lin Sun, Jiale Cao, Jin Xie, Fahad Shahbaz Khan, Yanwei Pang
Abstract: Stable diffusion has demonstrated strong image synthesis ability to given text descriptions, suggesting it to contain strong semantic clue for grouping objects. Inspired by this, researchers have explored employing stable diffusion for trainingfree segmentation. Most existing approaches either simply employ cross-attention map or refine it by self-attention map, to generate segmentation masks. We believe that iterative refinement with self-attention map would lead to better results. However, we mpirically demonstrate that such a refinement is sub-optimal likely due to the self-attention map containing irrelevant global information which hampers accurately refining cross-attention map with multiple iterations. To address this, we propose an iterative refinement framework for training-free segmentation, named iSeg, having an entropy-reduced self-attention module which utilizes a gradient descent scheme to reduce the entropy of self-attention map, thereby suppressing the weak responses corresponding to irrelevant global information. Leveraging the entropy-reduced self-attention module, our iSeg stably improves refined crossattention map with iterative refinement. Further, we design a category-enhanced cross-attention module to generate accurate cross-attention map, providing a better initial input for iterative refinement. Extensive experiments across different datasets and diverse segmentation tasks reveal the merits of proposed contributions, leading to promising performance on diverse segmentation tasks. For unsupervised semantic segmentation on Cityscapes, our iSeg achieves an absolute gain of 3.8% in terms of mIoU compared to the best existing training-free approach in literature. Moreover, our proposed iSeg can support segmentation with different kind of images and interactions.
Authors: Lulin Li, Ben Chen, Xuechao Zou, Junliang Xing, Pin Tao
Abstract: Owing to the diverse geographical environments, intricate landscapes, and high-density settlements, the automatic identification of urban village boundaries using remote sensing images is a highly challenging task. This paper proposes a novel and efficient neural network model called UV-Mamba for accurate boundary detection in high-resolution remote sensing images. UV-Mamba mitigates the memory loss problem in long sequence modeling, which arises in state space model (SSM) with increasing image size, by incorporating deformable convolutions (DCN). Its architecture utilizes an encoder-decoder framework, includes an encoder with four deformable state space augmentation (DSSA) blocks for efficient multi-level semantic extraction and a decoder to integrate the extracted semantic information. We conducted experiments on the Beijing and Xi'an datasets, and the results show that UV-Mamba achieves state-of-the-art performance. Specifically, our model achieves 73.3% and 78.1% IoU on the Beijing and Xi'an datasets, respectively, representing improvements of 1.2% and 3.4% IoU over the previous best model, while also being 6x faster in inference speed and 40x smaller in parameter count. Source code and pre-trained models are available in the supplementary material.
Authors: Zihong Sun, Hong Wang, Qi Xie, Yefeng Zheng, Deyu Meng
Abstract: Retinal vessel segmentation is of great clinical significance for the diagnosis of many eye-related diseases, but it is still a formidable challenge due to the intricate vascular morphology. With the skillful characterization of the translation symmetry existing in retinal vessels, convolutional neural networks (CNNs) have achieved great success in retinal vessel segmentation. However, the rotation-and-scale symmetry, as a more widespread image prior in retinal vessels, fails to be characterized by CNNs. Therefore, we propose a rotation-and-scale equivariant Fourier parameterized convolution (RSF-Conv) specifically for retinal vessel segmentation, and provide the corresponding equivariance analysis. As a general module, RSF-Conv can be integrated into existing networks in a plug-and-play manner while significantly reducing the number of parameters. For instance, we replace the traditional convolution filters in U-Net and Iter-Net with RSF-Convs, and faithfully conduct comprehensive experiments. RSF-Conv+U-Net and RSF-Conv+Iter-Net not only have slight advantages under in-domain evaluation, but more importantly, outperform all comparison methods by a significant margin under out-of-domain evaluation. It indicates the remarkable generalization of RSF-Conv, which holds greater practical clinical significance for the prevalent cross-device and cross-hospital challenges in clinical practice. To comprehensively demonstrate the effectiveness of RSF-Conv, we also apply RSF-Conv+U-Net and RSF-Conv+Iter-Net to retinal artery/vein classification and achieve promising performance as well, indicating its clinical application potential.
Authors: Yaodong Yu, Sam Buchanan, Druv Pai, Tianzhe Chu, Ziyang Wu, Shengbang Tong, Hao Bai, Yuexiang Zhai, Benjamin D. Haeffele, Yi Ma
Abstract: In this paper, we contend that a natural objective of representation learning is to compress and transform the distribution of the data, say sets of tokens, towards a low-dimensional Gaussian mixture supported on incoherent subspaces. The goodness of such a representation can be evaluated by a principled measure, called sparse rate reduction, that simultaneously maximizes the intrinsic information gain and extrinsic sparsity of the learned representation. From this perspective, popular deep network architectures, including transformers, can be viewed as realizing iterative schemes to optimize this measure. Particularly, we derive a transformer block from alternating optimization on parts of this objective: the multi-head self-attention operator compresses the representation by implementing an approximate gradient descent step on the coding rate of the features, and the subsequent multi-layer perceptron sparsifies the features. This leads to a family of white-box transformer-like deep network architectures, named CRATE, which are mathematically fully interpretable. We show, by way of a novel connection between denoising and compression, that the inverse to the aforementioned compressive encoding can be realized by the same class of CRATE architectures. Thus, the so-derived white-box architectures are universal to both encoders and decoders. Experiments show that these networks, despite their simplicity, indeed learn to compress and sparsify representations of large-scale real-world image and text datasets, and achieve performance very close to highly engineered transformer-based models: ViT, MAE, DINO, BERT, and GPT2. We believe the proposed computational framework demonstrates great potential in bridging the gap between theory and practice of deep learning, from a unified perspective of data compression. Code is available at: https://ma-lab-berkeley.github.io/CRATE .
Authors: Matthijs Douze, Alexandr Guzhva, Chengqi Deng, Jeff Johnson, Gergely Szilvasy, Pierre-Emmanuel Mazar\'e, Maria Lomeli, Lucas Hosseini, Herv\'e J\'egou
Abstract: Vector databases typically manage large collections of embedding vectors. Currently, AI applications are growing rapidly, and so is the number of embeddings that need to be stored and indexed. The Faiss library is dedicated to vector similarity search, a core functionality of vector databases. Faiss is a toolkit of indexing methods and related primitives used to search, cluster, compress and transform vectors. This paper describes the trade-off space of vector search and the design principles of Faiss in terms of structure, approach to optimization and interfacing. We benchmark key features of the library and discuss a few selected applications to highlight its broad applicability.
Authors: Yiting Qu, Xinyue Shen, Yixin Wu, Michael Backes, Savvas Zannettou, Yang Zhang
Abstract: With the advent of text-to-image models and concerns about their misuse, developers are increasingly relying on image safety classifiers to moderate their generated unsafe images. Yet, the performance of current image safety classifiers remains unknown for both real-world and AI-generated images. In this work, we propose UnsafeBench, a benchmarking framework that evaluates the effectiveness and robustness of image safety classifiers, with a particular focus on the impact of AI-generated images on their performance. First, we curate a large dataset of 10K real-world and AI-generated images that are annotated as safe or unsafe based on a set of 11 unsafe categories of images (sexual, violent, hateful, etc.). Then, we evaluate the effectiveness and robustness of five popular image safety classifiers, as well as three classifiers that are powered by general-purpose visual language models. Our assessment indicates that existing image safety classifiers are not comprehensive and effective enough to mitigate the multifaceted problem of unsafe images. Also, there exists a distribution shift between real-world and AI-generated images in image qualities, styles, and layouts, leading to degraded effectiveness and robustness. Motivated by these findings, we build a comprehensive image moderation tool called PerspectiveVision, which addresses the main drawbacks of existing classifiers with improved effectiveness and robustness, especially on AI-generated images. UnsafeBench and PerspectiveVision can aid the research community in better understanding the landscape of image safety classification in the era of generative AI.
Authors: Tianhao Peng, Chen Feng, Duolikun Danier, Fan Zhang, Benoit Vallade, Alex Mackin, David Bull
Abstract: With recent advances in deep learning, numerous algorithms have been developed to enhance video quality, reduce visual artifacts, and improve perceptual quality. However, little research has been reported on the quality assessment of enhanced content - the evaluation of enhancement methods is often based on quality metrics that were designed for compression applications. In this paper, we propose a novel blind deep video quality assessment (VQA) method specifically for enhanced video content. It employs a new Recurrent Memory Transformer (RMT) based network architecture to obtain video quality representations, which is optimized through a novel content-quality-aware contrastive learning strategy based on a new database containing 13K training patches with enhanced content. The extracted quality representations are then combined through linear regression to generate video-level quality indices. The proposed method, RMT-BVQA, has been evaluated on the VDPVE (VQA Dataset for Perceptual Video Enhancement) database through a five-fold cross validation. The results show its superior correlation performance when compared to ten existing no-reference quality metrics.
Authors: Zedian Shao, Hongbin Liu, Yuepeng Hu, Neil Zhenqiang Gong
Abstract: Multimodal large language models (MLLMs) have become the cornerstone of today's generative AI ecosystem, sparking intense competition among tech giants and startups. In particular, an MLLM generates a text response given a prompt consisting of an image and a question. While state-of-the-art MLLMs use safety filters and alignment techniques to refuse unsafe prompts, in this work, we introduce MLLM-Refusal, the first method that induces refusals for safe prompts. In particular, our MLLM-Refusal optimizes a nearly-imperceptible refusal perturbation and adds it to an image, causing target MLLMs to likely refuse a safe prompt containing the perturbed image and a safe question. Specifically, we formulate MLLM-Refusal as a constrained optimization problem and propose an algorithm to solve it. Our method offers competitive advantages for MLLM model providers by potentially disrupting user experiences of competing MLLMs, since competing MLLM's users will receive unexpected refusals when they unwittingly use these perturbed images in their prompts. We evaluate MLLM-Refusal on four MLLMs across four datasets, demonstrating its effectiveness in causing competing MLLMs to refuse safe prompts while not affecting non-competing MLLMs. Furthermore, we explore three potential countermeasures-adding Gaussian noise, DiffPure, and adversarial training. Our results show that though they can mitigate MLLM-Refusal's effectiveness, they also sacrifice the accuracy and/or efficiency of the competing MLLM. The code is available at https://github.com/Sadcardation/MLLM-Refusal.
Authors: Guido Maria D'Amely di Melendugno, Alessandro Flaborea, Pascal Mettes, Fabio Galasso
Abstract: Autonomous robots are increasingly becoming a strong fixture in social environments. Effective crowd navigation requires not only safe yet fast planning, but should also enable interpretability and computational efficiency for working in real-time on embedded devices. In this work, we advocate for hyperbolic learning to enable crowd navigation and we introduce Hyp2Nav. Different from conventional reinforcement learning-based crowd navigation methods, Hyp2Nav leverages the intrinsic properties of hyperbolic geometry to better encode the hierarchical nature of decision-making processes in navigation tasks. We propose a hyperbolic policy model and a hyperbolic curiosity module that results in effective social navigation, best success rates, and returns across multiple simulation settings, using up to 6 times fewer parameters than competitor state-of-the-art models. With our approach, it becomes even possible to obtain policies that work in 2-dimensional embedding spaces, opening up new possibilities for low-resource crowd navigation and model interpretability. Insightfully, the internal hyperbolic representation of Hyp2Nav correlates with how much attention the robot pays to the surrounding crowds, e.g. due to multiple people occluding its pathway or to a few of them showing colliding plans, rather than to its own planned route. The code is available at https://github.com/GDam90/hyp2nav.
Authors: Van Phi Nguyen, Tri Nhan Luong Ha, Huy Hieu Pham, Quoc Long Tran
Abstract: Conditional video diffusion models (CDM) have shown promising results for video synthesis, potentially enabling the generation of realistic echocardiograms to address the problem of data scarcity. However, current CDMs require a paired segmentation map and echocardiogram dataset. We present a new method called Free-Echo for generating realistic echocardiograms from a single end-diastolic segmentation map without additional training data. Our method is based on the 3D-Unet with Temporal Attention Layers model and is conditioned on the segmentation map using a training-free conditioning method based on SDEdit. We evaluate our model on two public echocardiogram datasets, CAMUS and EchoNet-Dynamic. We show that our model can generate plausible echocardiograms that are spatially aligned with the input segmentation map, achieving performance comparable to training-based CDMs. Our work opens up new possibilities for generating echocardiograms from a single segmentation map, which can be used for data augmentation, domain adaptation, and other applications in medical imaging. Our code is available at \url{https://github.com/gungui98/echo-free}
Authors: Jian Li, Weiheng Lu, Hao Fei, Meng Luo, Ming Dai, Min Xia, Yizhang Jin, Zhenye Gan, Ding Qi, Chaoyou Fu, Ying Tai, Wankou Yang, Yabiao Wang, Chengjie Wang
Abstract: Multimodal Large Language Models (MLLMs) are gaining increasing popularity in both academia and industry due to their remarkable performance in various applications such as visual question answering, visual perception, understanding, and reasoning. Over the past few years, significant efforts have been made to examine MLLMs from multiple perspectives. This paper presents a comprehensive review of 200 benchmarks and evaluations for MLLMs, focusing on (1)perception and understanding, (2)cognition and reasoning, (3)specific domains, (4)key capabilities, and (5)other modalities. Finally, we discuss the limitations of the current evaluation methods for MLLMs and explore promising future directions. Our key argument is that evaluation should be regarded as a crucial discipline to support the development of MLLMs better. For more details, please visit our GitHub repository: https://github.com/swordlidev/Evaluation-Multimodal-LLMs-Survey.
URLs: https://github.com/swordlidev/Evaluation-Multimodal-LLMs-Survey.