Authors: NVSL Narasimham, Keshav Kumar K
Brain Tumours (BT) are extremely dangerous and difficult to treat. Currently, doctors must manually examine images and manually mark out tumour regions to diagnose BT; this process is time-consuming and error-prone. In recent times, experts have proposed automating approaches for detecting BT at an early stage. The poor accuracy and highly incorrect prediction results of these methods caused them to start the research. In this study, we suggest a fuzzy logic-based system for categorising BT. This study used a dataset of 253 Magnetic Resonance Imaging (MRI) brain images that included tumour and healthy images. The images were first pre-processed. After that, we pull out features like tumour size and the image's global threshold value. The watershed and region-growing approach is used to calculate the tumour size. After that, the fuzzy system receives the two features as input. Accuracy, F1-score, precision, and recall are used to assess the results of the fuzzy by employing both size determination approaches. With the size input variable discovered by the region growth method and global threshold values, the fuzzy system outperforms the watershed method. The significance of this research lies in its potential to revolutionize brain tumour diagnosis by offering a more accurate and efficient automated classification system. By reducing human intervention and providing reliable results, this approach could assist medical professionals in making timely and precise decisions, leading to improved patient outcomes and potentially saving lives. The advancement of such automated techniques has the potential to pave the way for enhanced medical imaging analysis and, ultimately, better management of brain tumour cases.
Authors: Jon McCormack, Maria Teresa Llano, Stephen James Krol, Nina Rajcic
Image generation using generative AI is rapidly becoming a major new source of visual media, with billions of AI generated images created using diffusion models such as Stable Diffusion and Midjourney over the last few years. In this paper we collect and analyse over 3 million prompts and the images they generate. Using natural language processing, topic analysis and visualisation methods we aim to understand collectively how people are using text prompts, the impact of these systems on artists, and more broadly on the visual cultures they promote. Our study shows that prompting focuses largely on surface aesthetics, reinforcing cultural norms, popular conventional representations and imagery. We also find that many users focus on popular topics (such as making colouring books, fantasy art, or Christmas cards), suggesting that the dominant use for the systems analysed is recreational rather than artistic.
Authors: Caroline Mazini Rodrigues (LRDE, LIGM), Nicolas Boutry (LRDE), Laurent Najman (LIGM)
The explication of Convolutional Neural Networks (CNN) through xAI techniques often poses challenges in interpretation. The inherent complexity of input features, notably pixels extracted from images, engenders complex correlations. Gradient-based methodologies, exemplified by Integrated Gradients (IG), effectively demonstrate the significance of these features. Nevertheless, the conversion of these explanations into images frequently yields considerable noise. Presently, we introduce GAD (Gradient Artificial Distancing) as a supportive framework for gradient-based techniques. Its primary objective is to accentuate influential regions by establishing distinctions between classes. The essence of GAD is to limit the scope of analysis during visualization and, consequently reduce image noise. Empirical investigations involving occluded images have demonstrated that the identified regions through this methodology indeed play a pivotal role in facilitating class differentiation.
Authors: Nicolae Catalin Ristea, Ando Saabas, Ross Cutler, Babak Naderi, Sebastian Braun, Solomiya Branets
The ICASSP 2024 Speech Signal Improvement Grand Challenge is intended to stimulate research in the area of improving the speech signal quality in communication systems. This marks our second challenge, building upon the success from the previous ICASSP 2023 Grand Challenge. We enhance the competition by introducing a dataset synthesizer, enabling all participating teams to start at a higher baseline, an objective metric for our extended P.804 tests, transcripts for the 2023 test set, and we add Word Accuracy (WAcc) as a metric. We evaluate a total of 13 systems in the real-time track and 11 systems in the non-real-time track using both subjective P.804 and objective Word Accuracy metrics.
Authors: Zahra Babaiee, Peyman M. Kiasari, Daniela Rus, Radu Grosu
Recent advances in depthwise-separable convolutional neural networks (DS-CNNs) have led to novel architectures, that surpass the performance of classical CNNs, by a considerable scalability and accuracy margin. This paper reveals another striking property of DS-CNN architectures: discernible and explainable patterns emerge in their trained depthwise convolutional kernels in all layers. Through an extensive analysis of millions of trained filters, with different sizes and from various models, we employed unsupervised clustering with autoencoders, to categorize these filters. Astonishingly, the patterns converged into a few main clusters, each resembling the difference of Gaussian (DoG) functions, and their first and second-order derivatives. Notably, we were able to classify over 95\% and 90\% of the filters from state-of-the-art ConvNextV2 and ConvNeXt models, respectively. This finding is not merely a technological curiosity; it echoes the foundational models neuroscientists have long proposed for the vision systems of mammals. Our results thus deepen our understanding of the emergent properties of trained DS-CNNs and provide a bridge between artificial and biological visual processing systems. More broadly, they pave the way for more interpretable and biologically-inspired neural network designs in the future.
Authors: Muhammad Ahmed Chaudhry, Lyna Kim, Jeremy Irvin, Yuzu Ido, Sonia Chu, Jared Thomas Isobe, Andrew Y. Ng, Duncan Watson-Parris
Clouds play a significant role in global temperature regulation through their effect on planetary albedo. Anthropogenic emissions of aerosols can alter the albedo of clouds, but the extent of this effect, and its consequent impact on temperature change, remains uncertain. Human-induced clouds caused by ship aerosol emissions, commonly referred to as ship tracks, provide visible manifestations of this effect distinct from adjacent cloud regions and therefore serve as a useful sandbox to study human-induced clouds. However, the lack of large-scale ship track data makes it difficult to deduce their general effects on cloud formation. Towards developing automated approaches to localize ship tracks at scale, we present CloudTracks, a dataset containing 3,560 satellite images labeled with more than 12,000 ship track instance annotations. We train semantic segmentation and instance segmentation model baselines on our dataset and find that our best model substantially outperforms previous state-of-the-art for ship track localization (61.29 vs. 48.65 IoU). We also find that the best instance segmentation model is able to identify the number of ship tracks in each image more accurately than the previous state-of-the-art (1.64 vs. 4.99 MAE). However, we identify cases where the best model struggles to accurately localize and count ship tracks, so we believe CloudTracks will stimulate novel machine learning approaches to better detect elongated and overlapping features in satellite images. We release our dataset openly at {zenodo.org/records/10042922}.
Authors: Balamurali Murugesan, Sukesh Adiga Vasudeva, Bingyuan Liu, Hervé Lombaert, Ismail Ben Ayed, Jose Dolz
Ensuring reliable confidence scores from deep neural networks is of paramount significance in critical decision-making systems, particularly in real-world domains such as healthcare. Recent literature on calibrating deep segmentation networks has resulted in substantial progress. Nevertheless, these approaches are strongly inspired by the advancements in classification tasks, and thus their uncertainty is usually modeled by leveraging the information of individual pixels, disregarding the local structure of the object of interest. Indeed, only the recent Spatially Varying Label Smoothing (SVLS) approach considers pixel spatial relationships across classes, by softening the pixel label assignments with a discrete spatial Gaussian kernel. In this work, we first present a constrained optimization perspective of SVLS and demonstrate that it enforces an implicit constraint on soft class proportions of surrounding pixels. Furthermore, our analysis shows that SVLS lacks a mechanism to balance the contribution of the constraint with the primary objective, potentially hindering the optimization process. Based on these observations, we propose NACL (Neighbor Aware CaLibration), a principled and simple solution based on equality constraints on the logit values, which enables to control explicitly both the enforced constraint and the weight of the penalty, offering more flexibility. Comprehensive experiments on a wide variety of well-known segmentation benchmarks demonstrate the superior calibration performance of the proposed approach, without affecting its discriminative power. Furthermore, ablation studies empirically show the model agnostic nature of our approach, which can be used to train a wide span of deep segmentation networks.
Authors: Kumar Abhishek, Aditi Jain, Ghassan Hamarneh
The remarkable progress of deep learning in dermatological tasks has brought us closer to achieving diagnostic accuracies comparable to those of human experts. However, while large datasets play a crucial role in the development of reliable deep neural network models, the quality of data therein and their correct usage are of paramount importance. Several factors can impact data quality, such as the presence of duplicates, data leakage across train-test partitions, mislabeled images, and the absence of a well-defined test partition. In this paper, we conduct meticulous analyses of two popular dermatological image datasets: DermaMNIST and Fitzpatrick17k, uncovering these data quality issues, measure the effects of these problems on the benchmark results, and propose corrections to the datasets. Besides ensuring the reproducibility of our analysis, by making our analysis pipeline and the accompanying code publicly available, we aim to encourage similar explorations and to facilitate the identification and addressing of potential data quality issues in other large datasets.
Authors: Saumya Saxena, Mohit Sharma, Oliver Kroemer
Leveraging sensing modalities across diverse spatial and temporal resolutions can improve performance of robotic manipulation tasks. Multi-spatial resolution sensing provides hierarchical information captured at different spatial scales and enables both coarse and precise motions. Simultaneously multi-temporal resolution sensing enables the agent to exhibit high reactivity and real-time control. In this work, we propose a framework, MResT (Multi-Resolution Transformer), for learning generalizable language-conditioned multi-task policies that utilize sensing at different spatial and temporal resolutions using networks of varying capacities to effectively perform real time control of precise and reactive tasks. We leverage off-the-shelf pretrained vision-language models to operate on low-frequency global features along with small non-pretrained models to adapt to high frequency local feedback. Through extensive experiments in 3 domains (coarse, precise and dynamic manipulation tasks), we show that our approach significantly improves (2X on average) over recent multi-task baselines. Further, our approach generalizes well to visual and geometric variations in target objects and to varying interaction forces.
Authors: Fouad Afiouni, Mohamad Fakih, Joey Sleiman
Augmented Reality (AR) applications necessitates methods of inserting needed objects into scenes captured by cameras in a way that is coherent with the surroundings. Common AR applications require the insertion of predefined 3D objects with known properties and shape. This simplifies the problem since it is reduced to extracting an illumination model for the object in that scene by understanding the surrounding light sources. However, it is often not the case that we have information about the properties of an object, especially when we depart from a single source image. Our method renders such source fragments in a coherent way with the target surroundings using only these two images. Our pipeline uses a Deep Image Prior (DIP) network based on a U-Net architecture as the main renderer, alongside robust-feature extracting networks that are used to apply needed losses. Our method does not require any pair-labeled data, and no extensive training on a dataset. We compare our method using qualitative metrics to the baseline methods such as Cut and Paste, Cut And Paste Neural Rendering, and Image Harmonization
Authors: Guangyi Chen, Yifan Shen, Zhenhao Chen, Xiangchen Song, Yuewen Sun, Weiran Yao, Xiao Liu, Kun Zhang
Identifying the underlying time-delayed latent causal processes in sequential data is vital for grasping temporal dynamics and making downstream reasoning. While some recent methods can robustly identify these latent causal variables, they rely on strict assumptions about the invertible generation process from latent variables to observed data. However, these assumptions are often hard to satisfy in real-world applications containing information loss. For instance, the visual perception process translates a 3D space into 2D images, or the phenomenon of persistence of vision incorporates historical data into current perceptions. To address this challenge, we establish an identifiability theory that allows for the recovery of independent latent components even when they come from a nonlinear and non-invertible mix. Using this theory as a foundation, we propose a principled approach, CaRiNG, to learn the CAusal RepresentatIon of Non-invertible Generative temporal data with identifiability guarantees. Specifically, we utilize temporal context to recover lost latent information and apply the conditions in our theory to guide the training process. Through experiments conducted on synthetic datasets, we validate that our CaRiNG method reliably identifies the causal process, even when the generation process is non-invertible. Moreover, we demonstrate that our approach considerably improves temporal understanding and reasoning in practical applications.
Authors: Sanket Rajan Gupte, Josiah Aklilu, Jeffrey J. Nirschl, Serena Yeung-Levy
Foundation vision or vision-language models are trained on large unlabeled or noisy data and learn robust representations that can achieve impressive zero- or few-shot performance on diverse tasks. Given these properties, they are a natural fit for active learning (AL), which aims to maximize labeling efficiency, but the full potential of foundation models has not been explored in the context of AL, specifically in the low-budget regime. In this work, we evaluate how foundation models influence three critical components of effective AL, namely, 1) initial labeled pool selection, 2) ensuring diverse sampling, and 3) the trade-off between representative and uncertainty sampling. We systematically study how the robust representations of foundation models (DINOv2, OpenCLIP) challenge existing findings in active learning. Our observations inform the principled construction of a new simple and elegant AL strategy that balances uncertainty estimated via dropout with sample diversity. We extensively test our strategy on many challenging image classification benchmarks, including natural images as well as out-of-domain biomedical images that are relatively understudied in the AL literature. Source code will be made available.
Authors: Byoungsung Lim, Seong-Whan Lee
Recent advances in implicit function-based approaches have shown promising results in 3D human reconstruction from a single RGB image. However, these methods are not sufficient to extend to more general cases, often generating dragged or disconnected body parts, particularly for animated characters. We argue that these limitations stem from the use of the existing point-level 3D shape representation, which lacks holistic 3D context understanding. Voxel-based reconstruction methods are more suitable for capturing the entire 3D space at once, however, these methods are not practical for high-resolution reconstructions due to their excessive memory usage. To address these challenges, we introduce Tri-directional Implicit Function (TIFu), which is a vector-level representation that increases global 3D consistencies while significantly reducing memory usage compared to voxel representations. We also introduce a new algorithm in 3D reconstruction at an arbitrary resolution by aggregating vectors along three orthogonal axes, resolving inherent problems with regressing fixed dimension of vectors. Our approach achieves state-of-the-art performances in both our self-curated character dataset and the benchmark 3D human dataset. We provide both quantitative and qualitative analyses to support our findings.
Recognizing food images presents unique challenges due to the variable spatial layout and shape changes of ingredients with different cooking and cutting methods. This study introduces an advanced approach for recognizing ingredients segmented from food images. The method localizes the candidate regions of the ingredients using the locating and sliding window techniques. Then, these regions are assigned into ingredient classes using a CNN (Convolutional Neural Network)-based single-ingredient classification model trained on a dataset of single-ingredient images. To address the challenge of processing speed in multi-ingredient recognition, a novel model pruning method is proposed that enhances the efficiency of the classification model. Subsequently, the multi-ingredient identification is achieved through a decision-making scheme, incorporating two novel algorithms. The single-ingredient image dataset, designed in accordance with the book entitled "New Food Ingredients List FOODS 2021", encompasses 9982 images across 110 diverse categories, emphasizing variety in ingredient shapes. In addition, a multi-ingredient image dataset is developed to rigorously evaluate the performance of our approach. Experimental results validate the effectiveness of our method, particularly highlighting its improved capability in recognizing multiple ingredients. This marks a significant advancement in the field of food image analysis.
Authors: Hyeonwoo Cho, Chanmin Park, Jinyoung Kim, Won Hwa Kim
A domain shift occurs when training (source) and test (target) data diverge in their distribution. Test-time adaptation (TTA) addresses the domain shift problem, aiming to adopt a trained model on the source domain to the target domain in a scenario where only a well-trained source model and unlabeled target data are available. In this scenario, handling false labels in the target domain is crucial because they negatively impact the model performance. To deal with this problem, we propose to utilize cluster structure (i.e., {`Clean'} and {`Noisy'} regions within each cluster) in the target domain formulated by the source model. Given an initial clustering of target samples, we first partition clusters into {`Clean'} and {`Noisy'} regions defined based on cluster prototype (i.e., centroid of each cluster). As these regions have totally different distributions of the true pseudo-labels, we adopt distinct training strategies for the clean and noisy regions: we selectively train the target with clean pseudo-labels in the clean region, whereas we introduce mixup inputs representing intermediate features between clean and noisy regions to increase the compactness of the cluster. We conducted extensive experiments on multiple datasets in online/offline TTA settings, whose results demonstrate that our method, {CNA-TTA}, achieves state-of-the-art for most cases.
Authors: Tao He, Tongtong Wu, Dongyang Zhang, Guiduo Duan, Ke Qin, Yuan-Fang Li
Scene graph generation (SGG) endeavors to predict visual relationships between pairs of objects within an image. Prevailing SGG methods traditionally assume a one-off learning process for SGG. This conventional paradigm may necessitate repetitive training on all previously observed samples whenever new relationships emerge, mitigating the risk of forgetting previously acquired knowledge. This work seeks to address this pitfall inherent in a suite of prior relationship predictions. Motivated by the achievements of in-context learning in pretrained language models, our approach imbues the model with the capability to predict relationships and continuously acquire novel knowledge without succumbing to catastrophic forgetting. To achieve this goal, we introduce a novel and pragmatic framework for scene graph generation, namely Lifelong Scene Graph Generation (LSGG), where tasks, such as predicates, unfold in a streaming fashion. In this framework, the model is constrained to exclusive training on the present task, devoid of access to previously encountered training data, except for a limited number of exemplars, but the model is tasked with inferring all predicates it has encountered thus far. Rigorous experiments demonstrate the superiority of our proposed method over state-of-the-art SGG models in the context of LSGG across a diverse array of metrics. Besides, extensive experiments on the two mainstream benchmark datasets, VG and Open-Image(v6), show the superiority of our proposed model to a number of competitive SGG models in terms of continuous learning and conventional settings. Moreover, comprehensive ablation experiments demonstrate the effectiveness of each component in our model.
Authors: Wen Ma, Qiuwen Lou, Arman Kazemi, Julian Faraone, Tariq Afzal
Video quality can suffer from limited internet speed while being streamed by users. Compression artifacts start to appear when the bitrate decreases to match the available bandwidth. Existing algorithms either focus on removing the compression artifacts at the same video resolution, or on upscaling the video resolution but not removing the artifacts. Super resolution-only approaches will amplify the artifacts along with the details by default. We propose a lightweight convolutional neural network (CNN)-based algorithm which simultaneously performs artifacts reduction and super resolution (ARSR) by enhancing the feature extraction layers and designing a custom training dataset. The output of this neural network is evaluated for test streams compressed at low bitrates using variable bitrate (VBR) encoding. The output video quality shows a 4-6 increase in video multi-method assessment fusion (VMAF) score compared to traditional interpolation upscaling approaches such as Lanczos or Bicubic.
Authors: Ragib Amin Nihal, Benjamin Yen, Katsutoshi Itoyama, Kazuhiro Nakadai
The demand for accurate object detection in aerial imagery has surged with the widespread use of drones and satellite technology. Traditional object detection models, trained on datasets biased towards large objects, struggle to perform optimally in aerial scenarios where small, densely clustered objects are prevalent. To address this challenge, we present an innovative approach that combines super-resolution and an adapted lightweight YOLOv5 architecture. We employ a range of datasets, including VisDrone-2023, SeaDroneSee, VEDAI, and NWPU VHR-10, to evaluate our model's performance. Our Super Resolved YOLOv5 architecture features Transformer encoder blocks, allowing the model to capture global context and context information, leading to improved detection results, especially in high-density, occluded conditions. This lightweight model not only delivers improved accuracy but also ensures efficient resource utilization, making it well-suited for real-time applications. Our experimental results demonstrate the model's superior performance in detecting small and densely clustered objects, underlining the significance of dataset choice and architectural adaptation for this specific task. In particular, the method achieves 52.5% mAP on VisDrone, exceeding top prior works. This approach promises to significantly advance object detection in aerial imagery, contributing to more accurate and reliable results in a variety of real-world applications.
Authors: Kodai Kamiya, Toru Tamaki
We propose a new method for learning videos by aggregating multiple models by sequentially extracting video clips from untrimmed video. The proposed method reduces the correlation between clips by feeding clips to multiple models in turn and synchronizes these models through federated learning. Experimental results show that the proposed method improves the performance compared to the no synchronization.
Authors: Yanqi Ge, Ye Huang, Wen Li, Lixin Duan
We introduced SSR, which utilizes SAM (segment-anything) as a strong regularizer during training, to greatly enhance the robustness of the image encoder for handling various domains. Specifically, given the fact that SAM is pre-trained with a large number of images over the internet, which cover a diverse variety of domains, the feature encoding extracted by the SAM is obviously less dependent on specific domains when compared to the traditional ImageNet pre-trained image encoder. Meanwhile, the ImageNet pre-trained image encoder is still a mature choice of backbone for the semantic segmentation task, especially when the SAM is category-irrelevant. As a result, our SSR provides a simple yet highly effective design. It uses the ImageNet pre-trained image encoder as the backbone, and the intermediate feature of each stage (ie there are 4 stages in MiT-B5) is regularized by SAM during training. After extensive experimentation on GTA5$\rightarrow$Cityscapes, our SSR significantly improved performance over the baseline without introducing any extra inference overhead.
Authors: Oleksandr Fedoruk, Konrad Klimaszewski, Aleksander Ogonowski, Michał Kruk
The availability of training data is one of the main limitations in deep learning applications for medical imaging. Data augmentation is a popular approach to overcome this problem. A new approach is a Machine Learning based augmentation, in particular usage of Generative Adversarial Networks (GAN). In this case, GANs generate images similar to the original dataset so that the overall training data amount is bigger, which leads to better performance of trained networks. A GAN model consists of two networks, a generator and a discriminator interconnected in a feedback loop which creates a competitive environment. This work is a continuation of the previous research where we trained StyleGAN2-ADA by Nvidia on the limited COVID-19 chest X-ray image dataset. In this paper, we study the dependence of the GAN-based augmentation performance on dataset size with a focus on small samples. Two datasets are considered, one with 1000 images per class (4000 images in total) and the second with 500 images per class (2000 images in total). We train StyleGAN2-ADA with both sets and then, after validating the quality of generated images, we use trained GANs as one of the augmentations approaches in multi-class classification problems. We compare the quality of the GAN-based augmentation approach to two different approaches (classical augmentation and no augmentation at all) by employing transfer learning-based classification of COVID-19 chest X-ray images. The results are quantified using different classification quality metrics and compared to the results from the literature. The GAN-based augmentation approach is found to be comparable with classical augmentation in the case of medium and large datasets but underperforms in the case of smaller datasets. The correlation between the size of the original dataset and the quality of classification is visible independently from the augmentation approach.
Authors: Nuoyan Zhou, Dawei Zhou, Decheng Liu, Xinbo Gao, Nannan Wang
Deep neural networks are vulnerable to adversarial samples. Adversarial fine-tuning methods aim to enhance adversarial robustness through fine-tuning the naturally pre-trained model in an adversarial training manner. However, we identify that some latent features of adversarial samples are confused by adversarial perturbation and lead to an unexpectedly increasing gap between features in the last hidden layer of natural and adversarial samples. To address this issue, we propose a disentanglement-based approach to explicitly model and further remove the latent features that cause the feature gap. Specifically, we introduce a feature disentangler to separate out the latent features from the features of the adversarial samples, thereby boosting robustness by eliminating the latent features. Besides, we align features in the pre-trained model with features of adversarial samples in the fine-tuned model, to further benefit from the features from natural samples without confusion. Empirical evaluations on three benchmark datasets demonstrate that our approach surpasses existing adversarial fine-tuning methods and adversarial training baselines.
Authors: Ruibo Ming, Zhewei Huang, Zhuoxuan Ju, Jianming Hu, Lihui Peng, Shuchang Zhou
Video prediction, a fundamental task in computer vision, aims to enable models to generate sequences of future frames based on existing video content. This task has garnered widespread application across various domains. In this paper, we comprehensively survey both historical and contemporary works in this field, encompassing the most widely used datasets and algorithms. Our survey scrutinizes the challenges and evolving landscape of video prediction within the realm of computer vision. We propose a novel taxonomy centered on the stochastic nature of video prediction algorithms. This taxonomy accentuates the gradual transition from deterministic to generative prediction methodologies, underlining significant advancements and shifts in approach.
Authors: Sriram Reddy Mandhati, N. Lakmal Deshapriya, Chatura Lavanga Mendis, Kavinda Gunasekara, Frank Yrle, Angsana Chaksan, Sujit Sanjeev
Plastic pollution is a critical environmental issue, and detecting and monitoring plastic litter is crucial to mitigate its impact. This paper presents the methodology of mapping street-level litter, focusing primarily on plastic waste and the location of trash bins. Our methodology involves employing a deep learning technique to identify litter and trash bins from street-level imagery taken by a camera mounted on a vehicle. Subsequently, we utilized heat maps to visually represent the distribution of litter and trash bins throughout cities. Additionally, we provide details about the creation of an open-source dataset ("pLitterStreet") which was developed and utilized in our approach. The dataset contains more than 13,000 fully annotated images collected from vehicle-mounted cameras and includes bounding box labels. To evaluate the effectiveness of our dataset, we tested four well known state-of-the-art object detection algorithms (Faster R-CNN, RetinaNet, YOLOv3, and YOLOv5), achieving an average precision (AP) above 40%. While the results show average metrics, our experiments demonstrated the reliability of using vehicle-mounted cameras for plastic litter mapping. The "pLitterStreet" can also be a valuable resource for researchers and practitioners to develop and further improve existing machine learning models for detecting and mapping plastic litter in an urban environment. The dataset is open-source and more details about the dataset and trained models can be found at https://github.com/gicait/pLitter.
Authors: Zhenyu Bao, Guibiao Liao, Zhongyuan Zhao, Kanglin Liu, Qing Li, Guoping Qiu
Simultaneously achieving 3D reconstruction and new view synthesis for indoor environments has widespread applications but is technically very challenging. State-of-the-art methods based on implicit neural functions can achieve excellent 3D reconstruction results, but their performances on new view synthesis can be unsatisfactory. The exciting development of neural radiance field (NeRF) has revolutionized new view synthesis, however, NeRF-based models can fail to reconstruct clean geometric surfaces. We have developed a dual neural radiance field (Du-NeRF) to simultaneously achieve high-quality geometry reconstruction and view rendering. Du-NeRF contains two geometric fields, one derived from the SDF field to facilitate geometric reconstruction and the other derived from the density field to boost new view synthesis. One of the innovative features of Du-NeRF is that it decouples a view-independent component from the density field and uses it as a label to supervise the learning process of the SDF field. This reduces shape-radiance ambiguity and enables geometry and color to benefit from each other during the learning process. Extensive experiments demonstrate that Du-NeRF can significantly improve the performance of novel view synthesis and 3D reconstruction for indoor environments and it is particularly effective in constructing areas containing fine geometries that do not obey multi-view color consistency.
Authors: Chao Chen, Jie Liu, Chang Zhou, Jie Tang, Gangshan Wu
Lane detection is to determine the precise location and shape of lanes on the road. Despite efforts made by current methods, it remains a challenging task due to the complexity of real-world scenarios. Existing approaches, whether proposal-based or keypoint-based, suffer from depicting lanes effectively and efficiently. Proposal-based methods detect lanes by distinguishing and regressing a collection of proposals in a streamlined top-down way, yet lack sufficient flexibility in lane representation. Keypoint-based methods, on the other hand, construct lanes flexibly from local descriptors, which typically entail complicated post-processing. In this paper, we present a "Sketch-and-Refine" paradigm that utilizes the merits of both keypoint-based and proposal-based methods. The motivation is that local directions of lanes are semantically simple and clear. At the "Sketch" stage, local directions of keypoints can be easily estimated by fast convolutional layers. Then we can build a set of lane proposals accordingly with moderate accuracy. At the "Refine" stage, we further optimize these proposals via a novel Lane Segment Association Module (LSAM), which allows adaptive lane segment adjustment. Last but not least, we propose multi-level feature integration to enrich lane feature representations more efficiently. Based on the proposed "Sketch and Refine" paradigm, we propose a fast yet effective lane detector dubbed "SRLane". Experiments show that our SRLane can run at a fast speed (i.e., 278 FPS) while yielding an F1 score of 78.9\%. The source code is available at: https://github.com/passerer/SRLane.
Authors: Ayda Yurtoğlu, Sinan Sonlu, Yalım Doğan, Uğur Güdükbay
The successful portrayal of personality in digital characters improves communication and immersion. Current research focuses on expressing personality through modifying animations using heuristic rules or data-driven models. While studies suggest motion style highly influences the apparent personality, the role of appearance can be similarly essential. This work analyzes the influence of movement and appearance on the perceived personality of short videos altered by motion transfer networks. We label the personalities in conference video clips with a user study to determine the samples that best represent the Five-Factor model's high, neutral, and low traits. We alter these videos using the Thin-Plate Spline Motion Model, utilizing the selected samples as the source and driving inputs. We follow five different cases to study the influence of motion and appearance on personality perception. Our comparative study reveals that motion and appearance influence different factors: motion strongly affects perceived extraversion, and appearance helps convey agreeableness and neuroticism.
Authors: Vasiliy Usatyuk, Denis Sapozhnikov, Sergey Egorov
This paper presents a method for achieving equilibrium in the ISING Hamiltonian when confronted with unevenly distributed charges on an irregular grid. Employing (Multi-Edge) QC-LDPC codes and the Boltzmann machine, our approach involves dimensionally expanding the system, substituting charges with circulants, and representing distances through circulant shifts. This results in a systematic mapping of the charge system onto a space, transforming the irregular grid into a uniform configuration, applicable to Torical and Circular Hyperboloid Topologies. The paper covers fundamental definitions and notations related to QC-LDPC Codes, Multi-Edge QC-LDPC codes, and the Boltzmann machine. It explores the marginalization problem in code on the graph probabilistic models for evaluating the partition function, encompassing exact and approximate estimation techniques. Rigorous proof is provided for the attainability of equilibrium states for the Boltzmann machine under Torical and Circular Hyperboloid, paving the way for the application of our methodology. Practical applications of our approach are investigated in Finite Geometry QC-LDPC Codes, specifically in Material Science. The paper further explores its effectiveness in the realm of Natural Language Processing Transformer Deep Neural Networks, examining Generalized Repeat Accumulate Codes, Spatially-Coupled and Cage-Graph QC-LDPC Codes. The versatile and impactful nature of our topology-aware hardware-efficient quasi-cycle codes equilibrium method is showcased across diverse scientific domains without the use of specific section delineations.
Authors: Yuxiang Hui, Yang Liu, Yaofang Liu, Fan Jia, Jinshan Pan, Raymond Chan, Tieyong Zeng
Video restoration task aims to recover high-quality videos from low-quality observations. This contains various important sub-tasks, such as video denoising, deblurring and low-light enhancement, since video often faces different types of degradation, such as blur, low light, and noise. Even worse, these kinds of degradation could happen simultaneously when taking videos in extreme environments. This poses significant challenges if one wants to remove these artifacts at the same time. In this paper, to the best of our knowledge, we are the first to propose an efficient end-to-end video transformer approach for the joint task of video deblurring, low-light enhancement, and denoising. This work builds a novel multi-tier transformer where each tier uses a different level of degraded video as a target to learn the features of video effectively. Moreover, we carefully design a new tier-to-tier feature fusion scheme to learn video features incrementally and accelerate the training process with a suitable adaptive weighting scheme. We also provide a new Multiscene-Lowlight-Blur-Noise (MLBN) dataset, which is generated according to the characteristics of the joint task based on the RealBlur dataset and YouTube videos to simulate realistic scenes as far as possible. We have conducted extensive experiments, compared with many previous state-of-the-art methods, to show the effectiveness of our approach clearly.
Authors: Jon Alvarez Justo, Daniela Lupu, Milica Orlandic, Ion Necoara, Tor Arne Johansen
Hyperspectral Imaging comprises excessive data consequently leading to significant challenges for data processing, storage and transmission. Compressive Sensing has been used in the field of Hyperspectral Imaging as a technique to compress the large amount of data. This work addresses the recovery of hyperspectral images 2.5x compressed. A comparative study in terms of the accuracy and the performance of the convex FISTA/ADMM in addition to the greedy gOMP/BIHT/CoSaMP recovery algorithms is presented. The results indicate that the algorithms recover successfully the compressed data, yet the gOMP algorithm achieves superior accuracy and faster recovery in comparison to the other algorithms at the expense of high dependence on unknown sparsity level of the data to recover.
Authors: Yan Yang, Md Zakir Hossain, Xuesong Li, Shafin Rahman, Eric Stone
Spatial transcriptomics (ST) captures gene expression within distinct regions (i.e., windows) of a tissue slide. Traditional supervised learning frameworks applied to model ST are constrained to predicting expression from slide image windows for gene types seen during training, failing to generalize to unseen gene types. To overcome this limitation, we propose a semantic guided network (SGN), a pioneering zero-shot framework for predicting gene expression from slide image windows. Considering a gene type can be described by functionality and phenotype, we dynamically embed a gene type to a vector per its functionality and phenotype, and employ this vector to project slide image windows to gene expression in feature space, unleashing zero-shot expression prediction for unseen gene types. The gene type functionality and phenotype are queried with a carefully designed prompt from a pre-trained large language model (LLM). On standard benchmark datasets, we demonstrate competitive zero-shot performance compared to past state-of-the-art supervised learning approaches.
Authors: Hanz Cuevas-Velasquez, Charlie Hewitt, Sadegh Aliakbarian, Tadas Baltrušaitis
Our work addresses the problem of egocentric human pose estimation from downwards-facing cameras on head-mounted devices (HMD). This presents a challenging scenario, as parts of the body often fall outside of the image or are occluded. Previous solutions minimize this problem by using fish-eye camera lenses to capture a wider view, but these can present hardware design issues. They also predict 2D heat-maps per joint and lift them to 3D space to deal with self-occlusions, but this requires large network architectures which are impractical to deploy on resource-constrained HMDs. We predict pose from images captured with conventional rectilinear camera lenses. This resolves hardware design issues, but means body parts are often out of frame. As such, we directly regress probabilistic joint rotations represented as matrix Fisher distributions for a parameterized body model. This allows us to quantify pose uncertainties and explain out-of-frame or occluded joints. This also removes the need to compute 2D heat-maps and allows for simplified DNN architectures which require less compute. Given the lack of egocentric datasets using rectilinear camera lenses, we introduce the SynthEgo dataset, a synthetic dataset with 60K stereo images containing high diversity of pose, shape, clothing and skin tone. Our approach achieves state-of-the-art results for this challenging configuration, reducing mean per-joint position error by 23% overall and 58% for the lower body. Our architecture also has eight times fewer parameters and runs twice as fast as the current state-of-the-art. Experiments show that training on our synthetic dataset leads to good generalization to real world images without fine-tuning.
Authors: Jon Alvarez Justo, Milica Orlandic
Hyperspectral Imaging (HSI) is used in a wide range of applications such as remote sensing, yet the transmission of the HS images by communication data links becomes challenging due to the large number of spectral bands that the HS images contain together with the limited data bandwidth available in real applications. Compressive Sensing reduces the images by randomly subsampling the spectral bands of each spatial pixel and then it performs the image reconstruction of all the bands using recovery algorithms which impose sparsity in a certain transform domain. Since the image pixels are not strictly sparse, this work studies a data sparsification pre-processing stage prior to compression to ensure the sparsity of the pixels. The sparsified images are compressed $2.5\times$ and then recovered using the Generalized Orthogonal Matching Pursuit algorithm (gOMP) characterized by high accuracy, low computational requirements and fast convergence. The experiments are performed in five conventional hyperspectral images where the effect of different sparsification levels in the quality of the uncompressed as well as the recovered images is studied. It is concluded that the gOMP algorithm reconstructs the hyperspectral images with higher accuracy as well as faster convergence when the pixels are highly sparsified and hence at the expense of reducing the quality of the recovered images with respect to the original images.
Authors: Behrooz Razeghi, Parsa Rahimi, Sébastien Marcel
In this study, we harness the information-theoretic Privacy Funnel (PF) model to develop a method for privacy-preserving representation learning using an end-to-end training framework. We rigorously address the trade-off between obfuscation and utility. Both are quantified through the logarithmic loss, a measure also recognized as self-information loss. This exploration deepens the interplay between information-theoretic privacy and representation learning, offering substantive insights into data protection mechanisms for both discriminative and generative models. Importantly, we apply our model to state-of-the-art face recognition systems. The model demonstrates adaptability across diverse inputs, from raw facial images to both derived or refined embeddings, and is competent in tasks such as classification, reconstruction, and generation.
Authors: Songsong Tian, Lusi Li, Weijun Li, Hang Ran, Li Li, Xin Ning
Few-Shot Class-Incremental Learning (FSCIL) aims to enable deep neural networks to learn new tasks incrementally from a small number of labeled samples without forgetting previously learned tasks, closely mimicking human learning patterns. In this paper, we propose a novel approach called Prompt Learning for FSCIL (PL-FSCIL), which harnesses the power of prompts in conjunction with a pre-trained Vision Transformer (ViT) model to address the challenges of FSCIL effectively. Our work pioneers the use of visual prompts in FSCIL, which is characterized by its notable simplicity. PL-FSCIL consists of two distinct prompts: the Domain Prompt and the FSCIL Prompt. Both are vectors that augment the model by embedding themselves into the attention layer of the ViT model. Specifically, the Domain Prompt assists the ViT model in adapting to new data domains. The task-specific FSCIL Prompt, coupled with a prototype classifier, amplifies the model's ability to effectively handle FSCIL tasks. We validate the efficacy of PL-FSCIL on widely used benchmark datasets such as CIFAR-100 and CUB-200. The results showcase competitive performance, underscoring its promising potential for real-world applications where high-quality data is often scarce. The source code is available at: https://github.com/TianSongS/PL-FSCIL.
Authors: Jingyu Zhuang, Di Kang, Yan-Pei Cao, Guanbin Li, Liang Lin, Ying Shan
Text-driven 3D scene editing has gained significant attention owing to its convenience and user-friendliness. However, existing methods still lack accurate control of the specified appearance and location of the editing result due to the inherent limitations of the text description. To this end, we propose a 3D scene editing framework, TIPEditor, that accepts both text and image prompts and a 3D bounding box to specify the editing region. With the image prompt, users can conveniently specify the detailed appearance/style of the target content in complement to the text description, enabling accurate control of the appearance. Specifically, TIP-Editor employs a stepwise 2D personalization strategy to better learn the representation of the existing scene and the reference image, in which a localization loss is proposed to encourage correct object placement as specified by the bounding box. Additionally, TIPEditor utilizes explicit and flexible 3D Gaussian splatting as the 3D representation to facilitate local editing while keeping the background unchanged. Extensive experiments have demonstrated that TIP-Editor conducts accurate editing following the text and image prompts in the specified bounding box region, consistently outperforming the baselines in editing quality, and the alignment to the prompts, qualitatively and quantitatively.
Authors: Hubert Padusinski, Thilo Braun, Christian Steinhauser, Lennart Ries, Eric Sax
Are we heading for an iceberg with the current testing of machine vision? This work delves into the landscape of Machine Vision (MV) testing, which is heavily required in Highly Automated Driving (HAD) systems. Utilizing the metaphorical notion of navigating towards an iceberg, we discuss the potential shortcomings concealed within current testing strategies. We emphasize the urgent need for a deeper understanding of how to deal with the opaque functions of MV in development processes. As overlooked considerations can cost lives. Our main contribution is the hierarchical level model, which we call Granularity Grades. The model encourages a refined exploration of the multi-scaled depths of understanding about the circumstances of environments in which MV is intended to operate. This model aims to provide a holistic overview of all entities that may impact MV functions, ranging from relations of individual entities like object attributes to entire environmental scenes. The application of our model delivers a structured exploration of entities in a specific domain, their relationships and assigning results of a MV-under-test to construct an entity-relationship graph. Through clustering patterns of relations in the graph general MV deficits are arguable. In Summary, our work contributes to a more nuanced and systematized identification of deficits of a MV test object in correlation to holistic circumstances in HAD operating domains.
Authors: Shipeng Zhu, Pengfei Fang, Chenjie Zhu, Zuoyan Zhao, Qiang Xu, Hui Xue
Real-world text can be damaged by corrosion issues caused by environmental or human factors, which hinder the preservation of the complete styles of texts, e.g., texture and structure. These corrosion issues, such as graffiti signs and incomplete signatures, bring difficulties in understanding the texts, thereby posing significant challenges to downstream applications, e.g., scene text recognition and signature identification. Notably, current inpainting techniques often fail to adequately address this problem and have difficulties restoring accurate text images along with reasonable and consistent styles. Formulating this as an open problem of text image inpainting, this paper aims to build a benchmark to facilitate its study. In doing so, we establish two specific text inpainting datasets which contain scene text images and handwritten text images, respectively. Each of them includes images revamped by real-life and synthetic datasets, featuring pairs of original images, corrupted images, and other assistant information. On top of the datasets, we further develop a novel neural framework, Global Structure-guided Diffusion Model (GSDM), as a potential solution. Leveraging the global structure of the text as a prior, the proposed GSDM develops an efficient diffusion model to recover clean texts. The efficacy of our approach is demonstrated by thorough empirical study, including a substantial boost in both recognition accuracy and image quality. These findings not only highlight the effectiveness of our method but also underscore its potential to enhance the broader field of text image understanding and processing. Code and datasets are available at: https://github.com/blackprotoss/GSDM.
Authors: Dan Lin, Philip Hann Yung Lee, Yiming Li, Ruoyu Wang, Kim-Hui Yap, Bingbing Li, You Shing Ngim
Driver Action Recognition (DAR) is crucial in vehicle cabin monitoring systems. In real-world applications, it is common for vehicle cabins to be equipped with cameras featuring different modalities. However, multi-modality fusion strategies for the DAR task within car cabins have rarely been studied. In this paper, we propose a novel yet efficient multi-modality driver action recognition method based on dual feature shift, named DFS. DFS first integrates complementary features across modalities by performing modality feature interaction. Meanwhile, DFS achieves the neighbour feature propagation within single modalities, by feature shifting among temporal frames. To learn common patterns and improve model efficiency, DFS shares feature extracting stages among multiple modalities. Extensive experiments have been carried out to verify the effectiveness of the proposed DFS model on the Drive\&Act dataset. The results demonstrate that DFS achieves good performance and improves the efficiency of multi-modality driver action recognition.
Authors: Alessandro Baiocchi, Indro Spinelli, Alessandro Nicolosi, Simone Scardapane
The recent surge in 3D data acquisition has spurred the development of geometric deep learning models for point cloud processing, boosted by the remarkable success of transformers in natural language processing. While point cloud transformers (PTs) have achieved impressive results recently, their quadratic scaling with respect to the point cloud size poses a significant scalability challenge for real-world applications. To address this issue, we propose the Adaptive Point Cloud Transformer (AdaPT), a standard PT model augmented by an adaptive token selection mechanism. AdaPT dynamically reduces the number of tokens during inference, enabling efficient processing of large point clouds. Furthermore, we introduce a budget mechanism to flexibly adjust the computational cost of the model at inference time without the need for retraining or fine-tuning separate models. Our extensive experimental evaluation on point cloud classification tasks demonstrates that AdaPT significantly reduces computational complexity while maintaining competitive accuracy compared to standard PTs. The code for AdaPT is made publicly available.
Authors: Rui Qiao, Bryan Kian Hsiang Low
Despite the rapid development of machine learning algorithms for domain generalization (DG), there is no clear empirical evidence that the existing DG algorithms outperform the classic empirical risk minimization (ERM) across standard benchmarks. To better understand this phenomenon, we investigate whether there are benefits of DG algorithms over ERM through the lens of label noise. Specifically, our finite-sample analysis reveals that label noise exacerbates the effect of spurious correlations for ERM, undermining generalization. Conversely, we illustrate that DG algorithms exhibit implicit label-noise robustness during finite-sample training even when spurious correlation is present. Such desirable property helps mitigate spurious correlations and improve generalization in synthetic experiments. However, additional comprehensive experiments on real-world benchmark datasets indicate that label-noise robustness does not necessarily translate to better performance compared to ERM. We conjecture that the failure mode of ERM arising from spurious correlations may be less pronounced in practice.
Authors: Xinyao Yu, Hao Sun, Ziwei Niu, Rui Qin, Zhenjia Bai, Yen-Wei Chen, Lanfen Lin
In recent years, large-scale pre-trained multimodal models (LMM) generally emerge to integrate the vision and language modalities, achieving considerable success in various natural language processing and computer vision tasks. The growing size of LMMs, however, results in a significant computational cost for fine-tuning these models for downstream tasks. Hence, prompt-based interaction strategy is studied to align modalities more efficiently. In this contex, we propose a novel prompt-based multimodal interaction strategy inspired by human memory strategy, namely Memory-Inspired Temporal Prompt Interaction (MITP). Our proposed method involves in two stages as in human memory strategy: the acquiring stage, and the consolidation and activation stage. We utilize temporal prompts on intermediate layers to imitate the acquiring stage, leverage similarity-based prompt interaction to imitate memory consolidation, and employ prompt generation strategy to imitate memory activation. The main strength of our paper is that we interact the prompt vectors on intermediate layers to leverage sufficient information exchange between modalities, with compressed trainable parameters and memory usage. We achieve competitive results on several datasets with relatively small memory usage and 2.0M of trainable parameters (about 1% of the pre-trained foundation model).
Authors: Lingchen Yang, Byungsoo Kim, Gaspard Zoss, Baran Gözcü, Markus Gross, Barbara Solenthaler
Active soft bodies can affect their shape through an internal actuation mechanism that induces a deformation. Similar to recent work, this paper utilizes a differentiable, quasi-static, and physics-based simulation layer to optimize for actuation signals parameterized by neural networks. Our key contribution is a general and implicit formulation to control active soft bodies by defining a function that enables a continuous mapping from a spatial point in the material space to the actuation value. This property allows us to capture the signal's dominant frequencies, making the method discretization agnostic and widely applicable. We extend our implicit model to mandible kinematics for the particular case of facial animation and show that we can reliably reproduce facial expressions captured with high-quality capture systems. We apply the method to volumetric soft bodies, human poses, and facial expressions, demonstrating artist-friendly properties, such as simple control over the latent space and resolution invariance at test time.
Authors: Yu-Shan Tai, An-Yeu (Andy)Wu
While vision transformers (ViTs) have shown great potential in computer vision tasks, their intense computation and memory requirements pose challenges for practical applications. Existing post-training quantization methods leverage value redistribution or specialized quantizers to address the non-normal distribution in ViTs. However, without considering the asymmetry in activations and relying on hand-crafted settings, these methods often struggle to maintain performance under low-bit quantization. To overcome these challenges, we introduce SmoothQuant with bias term (SQ-b) to alleviate the asymmetry issue and reduce the clamping loss. We also introduce optimal scaling factor ratio search (OPT-m) to determine quantization parameters by a data-dependent mechanism automatically. To further enhance the compressibility, we incorporate the above-mentioned techniques and propose a mixed-precision post-training quantization framework for vision transformers (MPTQ-ViT). We develop greedy mixed-precision quantization (Greedy MP) to allocate layer-wise bit-width considering both model performance and compressibility. Our experiments on ViT, DeiT, and Swin demonstrate significant accuracy improvements compared with SOTA on the ImageNet dataset. Specifically, our proposed methods achieve accuracy improvements ranging from 0.90% to 23.35% on 4-bit ViTs with single-precision and from 3.82% to 78.14% on 5-bit fully quantized ViTs with mixed-precision.
Authors: Florian Kluger, Bodo Rosenhahn
We present a real-time method for robust estimation of multiple instances of geometric models from noisy data. Geometric models such as vanishing points, planar homographies or fundamental matrices are essential for 3D scene analysis. Previous approaches discover distinct model instances in an iterative manner, thus limiting their potential for speedup via parallel computation. In contrast, our method detects all model instances independently and in parallel. A neural network segments the input data into clusters representing potential model instances by predicting multiple sets of sample and inlier weights. Using the predicted weights, we determine the model parameters for each potential instance separately in a RANSAC-like fashion. We train the neural network via task-specific loss functions, i.e. we do not require a ground-truth segmentation of the input data. As suitable training data for homography and fundamental matrix fitting is scarce, we additionally present two new synthetic datasets. We demonstrate state-of-the-art performance on these as well as multiple established datasets, with inference times as small as five milliseconds per image.
Authors: Hanxiao Tan
In recent years, the performance of point cloud models has been rapidly improved. However, due to the limited amount of relevant explainability studies, the unreliability and opacity of these black-box models may lead to potential risks in applications where human lives are at stake, e.g. autonomous driving or healthcare. This work proposes a DDPM-based point cloud global explainability method (DAM) that leverages Point Diffusion Transformer (PDT), a novel point-wise symmetric model, with dual-classifier guidance to generate high-quality global explanations. In addition, an adapted path gradient integration method for DAM is proposed, which not only provides a global overview of the saliency maps for point cloud categories, but also sheds light on how the attributions of the explanations vary during the generation process. Extensive experiments indicate that our method outperforms existing ones in terms of perceptibility, representativeness, and diversity, with a significant reduction in generation time. Our code is available at: https://github.com/Explain3D/DAM
Authors: Shruthi Gowda, Bahram Zonooz, Elahe Arani
Adversarial training improves the robustness of neural networks against adversarial attacks, albeit at the expense of the trade-off between standard and robust generalization. To unveil the underlying factors driving this phenomenon, we examine the layer-wise learning capabilities of neural networks during the transition from a standard to an adversarial setting. Our empirical findings demonstrate that selectively updating specific layers while preserving others can substantially enhance the network's learning capacity. We therefore propose CURE, a novel training framework that leverages a gradient prominence criterion to perform selective conservation, updating, and revision of weights. Importantly, CURE is designed to be dataset- and architecture-agnostic, ensuring its applicability across various scenarios. It effectively tackles both memorization and overfitting issues, thus enhancing the trade-off between robustness and generalization and additionally, this training approach also aids in mitigating "robust overfitting". Furthermore, our study provides valuable insights into the mechanisms of selective adversarial training and offers a promising avenue for future research.
Authors: Xiaoxiao Ma, Zhixiang Wei, Yi Jin, Pengyang Ling, Tianle Liu, Ben Wang, Junkang Dai, Huaian Chen, Enhong Chen
In this work, we observe that the model, which is trained on vast general images using masking strategy, has been naturally embedded with the distribution knowledge regarding natural images, and thus spontaneously attains the underlying potential for strong image denoising. Based on this observation, we propose a novel zero-shot denoising paradigm, i.e., Masked Pre-train then Iterative fill (MPI). MPI pre-trains a model with masking and fine-tunes it for denoising of a single image with unseen noise degradation. Concretely, the proposed MPI comprises two key procedures: 1) Masked Pre-training involves training a model on multiple natural images with random masks to gather generalizable representations, allowing for practical applications in varying noise degradation and even in distinct image types. 2) Iterative filling is devised to efficiently fuse pre-trained knowledge for denoising. Similar to but distinct from pre-training, random masking is retained to bridge the gap, but only the predicted parts covered by masks are assembled for efficiency, which enables high-quality denoising within a limited number of iterations. Comprehensive experiments across various noisy scenarios underscore the notable advances of proposed MPI over previous approaches with a marked reduction in inference time. Code is available at https://github.com/krennic999/MPI.git.
Authors: Baoyuan Wu, Hongrui Chen, Mingda Zhang, Zihao Zhu, Shaokui Wei, Danni Yuan, Mingli Zhu, Ruotong Wang, Li Liu, Chao Shen
As an emerging and vital topic for studying deep neural networks' vulnerability (DNNs), backdoor learning has attracted increasing interest in recent years, and many seminal backdoor attack and defense algorithms are being developed successively or concurrently, in the status of a rapid arms race. However, mainly due to the diverse settings, and the difficulties of implementation and reproducibility of existing works, there is a lack of a unified and standardized benchmark of backdoor learning, causing unfair comparisons, and unreliable conclusions (e.g., misleading, biased or even false conclusions). Consequently, it is difficult to evaluate the current progress and design the future development roadmap of this literature. To alleviate this dilemma, we build a comprehensive benchmark of backdoor learning called BackdoorBench. Our benchmark makes three valuable contributions to the research community. 1) We provide an integrated implementation of state-of-the-art (SOTA) backdoor learning algorithms (currently including 16 attack and 27 defense algorithms), based on an extensible modular-based codebase. 2) We conduct comprehensive evaluations of 12 attacks against 16 defenses, with 5 poisoning ratios, based on 4 models and 4 datasets, thus 11,492 pairs of evaluations in total. 3) Based on above evaluations, we present abundant analysis from 8 perspectives via 18 useful analysis tools, and provide several inspiring insights about backdoor learning. We hope that our efforts could build a solid foundation of backdoor learning to facilitate researchers to investigate existing algorithms, develop more innovative algorithms, and explore the intrinsic mechanism of backdoor learning. Finally, we have created a user-friendly website at this http URL, which collects all important information of BackdoorBench, including codebase, docs, leaderboard, and model Zoo.
Authors: Jan-Philipp Redlich, Friedrich Feuerhake, Joachim Weis, Nadine S. Schaadt, Sarah Teuber-Hanselmann, Christoph Buck, Sabine Luttmann, Andrea Eberle, Stefan Nikolin, Arno Appenzeller, Andreas Portmann, André Homeyer
In recent years, the diagnosis of gliomas has become increasingly complex. Histological assessment of glioma tissue using modern machine learning techniques offers new opportunities to support diagnosis and outcome prediction. To give an overview of the current state of research, this review examines 70 publicly available research studies on machine learning-based analysis of stained human glioma tissue sections, covering the diagnostic tasks of subtyping (16/70), grading (23/70), molecular marker prediction (13/70), and survival prediction (27/70). All studies were reviewed with regard to methodological aspects as well as clinical applicability. It was found that the focus of current research is the assessment of hematoxylin and eosin-stained tissue sections of adult-type diffuse gliomas. The majority of studies (49/70) are based on the publicly available glioblastoma and low-grade glioma datasets from The Cancer Genome Atlas (TCGA) and only a few studies employed other datasets in isolation (10/70) or in addition to the TCGA datasets (11/70). Current approaches mostly rely on convolutional neural networks (53/70) for analyzing tissue at 20x magnification (30/70). A new field of research is the integration of clinical data, omics data, or magnetic resonance imaging (27/70). So far, machine learning-based methods have achieved promising results, but are not yet used in real clinical settings. Future work should focus on the independent validation of methods on larger, multi-site datasets with high-quality and up-to-date clinical and molecular pathology annotations to demonstrate routine applicability.
Authors: Juan Castorena
This work leverages neural radiance fields and remote sensing for forestry applications. Here, we show neural radiance fields offer a wide range of possibilities to improve upon existing remote sensing methods in forest monitoring. We present experiments that demonstrate their potential to: (1) express fine features of forest 3D structure, (2) fuse available remote sensing modalities and (3), improve upon 3D structure derived forest metrics. Altogether, these properties make neural fields an attractive computational tool with great potential to further advance the scalability and accuracy of forest monitoring programs.
Authors: Dmytro Zakharov, Oleksandr Kuznetsov, Emanuele Frontoni
In the realm of security applications, biometric authentication systems play a crucial role, yet one often encounters challenges concerning privacy and security while developing one. One of the most fundamental challenges lies in avoiding storing biometrics directly in the storage but still achieving decently high accuracy. Addressing this issue, we contribute to both artificial intelligence and engineering fields. We introduce an innovative image distortion technique that effectively renders facial images unrecognizable to the eye while maintaining their identifiability by neural network models. From the theoretical perspective, we explore how reliable state-of-the-art biometrics recognition neural networks are by checking the maximal degree of image distortion, which leaves the predicted identity unchanged. On the other hand, applying this technique demonstrates a practical solution to the engineering challenge of balancing security, precision, and performance in biometric authentication systems. Through experimenting on the widely used datasets, we assess the effectiveness of our method in preserving AI feature representation and distorting relative to conventional metrics. We also compare our method with previously used approaches.
Authors: Mikael A. Mousse, Bethel C. A. R. K. Atohoun, Cina Motamed
Tracking ripening tomatoes is time consuming and labor intensive. Artificial intelligence technologies combined with those of computer vision can help users optimize the process of monitoring the ripening status of plants. To this end, we have proposed a tomato ripening monitoring approach based on deep learning in complex scenes. The objective is to detect mature tomatoes and harvest them in a timely manner. The proposed approach is declined in two parts. Firstly, the images of the scene are transmitted to the pre-processing layer. This process allows the detection of areas of interest (area of the image containing tomatoes). Then, these images are used as input to the maturity detection layer. This layer, based on a deep neural network learning algorithm, classifies the tomato thumbnails provided to it in one of the following five categories: green, brittle, pink, pale red, mature red. The experiments are based on images collected from the internet gathered through searches using tomato state across diverse languages including English, German, French, and Spanish. The experimental results of the maturity detection layer on a dataset composed of images of tomatoes taken under the extreme conditions, gave a good classification rate.
Authors: Chaochao Lu, Chen Qian, Guodong Zheng, Hongxing Fan, Hongzhi Gao, Jie Zhang, Jing Shao, Jingyi Deng, Jinlan Fu, Kexin Huang, Kunchang Li, Lijun Li, Limin Wang, Lu Sheng, Meiqi Chen, Ming Zhang, Qibing Ren, Sirui Chen, Tao Gui, Wanli Ouyang, Yali Wang, Yan Teng, Yaru Wang, Yi Wang, Yinan He, Yingchun Wang, Yixu Wang, Yongting Zhang, Yu Qiao, Yujiong Shen, Yurong Mou, Yuxi Chen, Zaibin Zhang, Zhelun Shi, Zhenfei Yin, Zhipin Wang
Multi-modal Large Language Models (MLLMs) have shown impressive abilities in generating reasonable responses with respect to multi-modal contents. However, there is still a wide gap between the performance of recent MLLM-based applications and the expectation of the broad public, even though the most powerful OpenAI's GPT-4 and Google's Gemini have been deployed. This paper strives to enhance understanding of the gap through the lens of a qualitative study on the generalizability, trustworthiness, and causal reasoning capabilities of recent proprietary and open-source MLLMs across four modalities: ie, text, code, image, and video, ultimately aiming to improve the transparency of MLLMs. We believe these properties are several representative factors that define the reliability of MLLMs, in supporting various downstream applications. To be specific, we evaluate the closed-source GPT-4 and Gemini and 6 open-source LLMs and MLLMs. Overall we evaluate 230 manually designed cases, where the qualitative results are then summarized into 12 scores (ie, 4 modalities times 3 properties). In total, we uncover 14 empirical findings that are useful to understand the capabilities and limitations of both proprietary and open-source MLLMs, towards more reliable downstream multi-modal applications.
Authors: Yue Yang, Atith N Gandhi, Greg Turk
Generative models such as GANs and diffusion models have demonstrated impressive image generation capabilities. Despite these successes, these systems are surprisingly poor at creating images with hands. We propose a novel training framework for generative models that substantially improves the ability of such systems to create hand images. Our approach is to augment the training images with three additional channels that provide annotations to hands in the image. These annotations provide additional structure that coax the generative model to produce higher quality hand images. We demonstrate this approach on two different generative models: a generative adversarial network and a diffusion model. We demonstrate our method both on a new synthetic dataset of hand images and also on real photographs that contain hands. We measure the improved quality of the generated hands through higher confidence in finger joint identification using an off-the-shelf hand detector.
Authors: Liangrui Pan, Lian Wang, Zhichao Feng, Zhujun Xu, Liwen Xu, Shaoliang Peng
Colorectal cancer (CRC) is among the top three malignant tumor types in terms of morbidity and mortality. Histopathological images are the gold standard for diagnosing colon cancer. Cellular nuclei instance segmentation and classification, and nuclear component regression tasks can aid in the analysis of the tumor microenvironment in colon tissue. Traditional methods are still unable to handle both types of tasks end-to-end at the same time, and have poor prediction accuracy and high application costs. This paper proposes a new UNet model for handling nuclei based on the UNet framework, called MGTUNet, which uses Mish, Group normalization and transposed convolution layer to improve the segmentation model, and a ranger optimizer to adjust the SmoothL1Loss values. Secondly, it uses different channels to segment and classify different types of nucleus, ultimately completing the nuclei instance segmentation and classification task, and the nuclei component regression task simultaneously. Finally, we did extensive comparison experiments using eight segmentation models. By comparing the three evaluation metrics and the parameter sizes of the models, MGTUNet obtained 0.6254 on PQ, 0.6359 on mPQ, and 0.8695 on R2. Thus, the experiments demonstrated that MGTUNet is now a state-of-the-art method for quantifying histopathological images of colon cancer.
Authors: Graham West, Matthew I. Swindall, Ben Keener, Timothy Player, Alex C. Williams, James H. Brusuelas, John F. Wallin
Performing classification on noisy, crowdsourced image datasets can prove challenging even for the best neural networks. Two issues which complicate the problem on such datasets are class imbalance and ground-truth uncertainty in labeling. The AL-ALL and AL-PUB datasets - consisting of tightly cropped, individual characters from images of ancient Greek papyri - are strongly affected by both issues. The application of ensemble modeling to such datasets can help identify images where the ground-truth is questionable and quantify the trustworthiness of those samples. As such, we apply stacked generalization consisting of nearly identical ResNets with different loss functions: one utilizing sparse cross-entropy (CXE) and the other Kullback-Liebler Divergence (KLD). Both networks use labels drawn from a crowd-sourced consensus. This consensus is derived from a Normalized Distribution of Annotations (NDA) based on all annotations for a given character in the dataset. For the second network, the KLD is calculated with respect to the NDA. For our ensemble model, we apply a k-nearest neighbors model to the outputs of the CXE and KLD networks. Individually, the ResNet models have approximately 93% accuracy, while the ensemble model achieves an accuracy of > 95%, increasing the classification trustworthiness. We also perform an analysis of the Shannon entropy of the various models' output distributions to measure classification uncertainty. Our results suggest that entropy is useful for predicting model misclassifications.
Authors: Hanwen Jiang, Zhenyu Jiang, Kristen Grauman, Yuke Zhu
While object reconstruction has made great strides in recent years, current methods typically require densely captured images and/or known camera poses, and generalize poorly to novel object categories. To step toward object reconstruction in the wild, this work explores reconstructing general real-world objects from a few images without known camera poses or object categories. The crux of our work is solving two fundamental 3D vision problems -- shape reconstruction and pose estimation -- in a unified approach. Our approach captures the synergies of these two problems: reliable camera pose estimation gives rise to accurate shape reconstruction, and the accurate reconstruction, in turn, induces robust correspondence between different views and facilitates pose estimation. Our method FORGE predicts 3D features from each view and leverages them in conjunction with the input images to establish cross-view correspondence for estimating relative camera poses. The 3D features are then transformed by the estimated poses into a shared space and are fused into a neural radiance field. The reconstruction results are rendered by volume rendering techniques, enabling us to train the model without 3D shape ground-truth. Our experiments show that FORGE reliably reconstructs objects from five views. Our pose estimation method outperforms existing ones by a large margin. The reconstruction results under predicted poses are comparable to the ones using ground-truth poses. The performance on novel testing categories matches the results on categories seen during training. Project page: https://ut-austin-rpl.github.io/FORGE/
Authors: Natalia Valderrama, Paola Ruiz Puentes, Isabela Hernández, Nicolás Ayobi, Mathilde Verlyk, Jessica Santander, Juan Caicedo, Nicolás Fernández, Pablo Arbeláez
Most benchmarks for studying surgical interventions focus on a specific challenge instead of leveraging the intrinsic complementarity among different tasks. In this work, we present a new experimental framework towards holistic surgical scene understanding. First, we introduce the Phase, Step, Instrument, and Atomic Visual Action recognition (PSI-AVA) Dataset. PSI-AVA includes annotations for both long-term (Phase and Step recognition) and short-term reasoning (Instrument detection and novel Atomic Action recognition) in robot-assisted radical prostatectomy videos. Second, we present Transformers for Action, Phase, Instrument, and steps Recognition (TAPIR) as a strong baseline for surgical scene understanding. TAPIR leverages our dataset's multi-level annotations as it benefits from the learned representation on the instrument detection task to improve its classification capacity. Our experimental results in both PSI-AVA and other publicly available databases demonstrate the adequacy of our framework to spur future research on holistic surgical scene understanding.
Authors: Baoliang Chen, Lingyu Zhu, Hanwei Zhu, Wenhan Yang, Linqi Song, Shiqi Wang
There is a growing consensus in the research community that the optimization of low-light image enhancement approaches should be guided by the visual quality perceived by end users. Despite the substantial efforts invested in the design of low-light enhancement algorithms, there has been comparatively limited focus on assessing subjective and objective quality systematically. To mitigate this gap and provide a clear path towards optimizing low-light image enhancement for better visual quality, we propose a gap-closing framework. In particular, our gap-closing framework starts with the creation of a large-scale dataset for Subjective QUality Assessment of REconstructed LOw-Light Images (SQUARE-LOL). This database serves as the foundation for studying the quality of enhanced images and conducting a comprehensive subjective user study. Subsequently, we propose an objective quality assessment measure that plays a critical role in bridging the gap between visual quality and enhancement. Finally, we demonstrate that our proposed objective quality measure can be incorporated into the process of optimizing the learning of the enhancement model toward perceptual optimality. We validate the effectiveness of our proposed framework through both the accuracy of quality prediction and the perceptual quality of image enhancement. Our database and code will be made publicly available to facilitate further research in this area.
Authors: Louisa Conwill, Samuel Anthony, Walter Scheirer
Augmented reality and other photo editing filters are popular methods used to modify faces online. Considering the important role of facial perception in communication, how do we perceive this increasing number of modified faces? In this paper we present the results of six surveys that measure familiarity with different styles of facial filters, perceived strangeness of faces edited with different filters, and ability to discern whether images are filtered. Our results demonstrate that faces modified with more traditional face filters are perceived similarly to unmodified faces, and faces filtered with augmented reality filters are perceived differently from unmodified faces. We discuss possible explanations for these results, including a societal adjustment to traditional photo editing techniques or the inherent differences in the different types of filters. We conclude with a discussion of how to build online spaces more responsibly based on our results.
Authors: Nicolás Ayobi, Alejandra Pérez-Rondón, Santiago Rodríguez, Pablo Arbeláez
We propose Masked-Attention Transformers for Surgical Instrument Segmentation (MATIS), a two-stage, fully transformer-based method that leverages modern pixel-wise attention mechanisms for instrument segmentation. MATIS exploits the instance-level nature of the task by employing a masked attention module that generates and classifies a set of fine instrument region proposals. Our method incorporates long-term video-level information through video transformers to improve temporal consistency and enhance mask classification. We validate our approach in the two standard public benchmarks, Endovis 2017 and Endovis 2018. Our experiments demonstrate that MATIS' per-frame baseline outperforms previous state-of-the-art methods and that including our temporal consistency module boosts our model's performance further.
Authors: Felix Ott, Lucas Heublein, David Rügamer, Bernd Bischl, Christopher Mutschler
The localization of objects is a crucial task in various applications such as robotics, virtual and augmented reality, and the transportation of goods in warehouses. Recent advances in deep learning have enabled the localization using monocular visual cameras. While structure from motion (SfM) predicts the absolute pose from a point cloud, absolute pose regression (APR) methods learn a semantic understanding of the environment through neural networks. However, both fields face challenges caused by the environment such as motion blur, lighting changes, repetitive patterns, and feature-less structures. This study aims to address these challenges by incorporating additional information and regularizing the absolute pose using relative pose regression (RPR) methods. RPR methods suffer under different challenges, i.e., motion blur. The optical flow between consecutive images is computed using the Lucas-Kanade algorithm, and the relative pose is predicted using an auxiliary small recurrent convolutional network. The fusion of absolute and relative poses is a complex task due to the mismatch between the global and local coordinate systems. State-of-the-art methods fusing absolute and relative poses use pose graph optimization (PGO) to regularize the absolute pose predictions using relative poses. In this work, we propose recurrent fusion networks to optimally align absolute and relative pose predictions to improve the absolute pose prediction. We evaluate eight different recurrent units and construct a simulation environment to pre-train the APR and RPR networks for better generalized training. Additionally, we record a large database of different scenarios in a challenging large-scale indoor environment that mimics a warehouse with transportation robots. We conduct hyperparameter searches and experiments to show the effectiveness of our recurrent fusion method compared to PGO.
Authors: Yanpeng Zhao, Siyu Gao, Yunbo Wang, Xiaokang Yang
Unsupervised learning of object-centric representations in dynamic visual scenes is challenging. Unlike most previous approaches that learn to decompose 2D images, we present DynaVol, a 3D scene generative model that unifies geometric structures and object-centric learning in a differentiable volume rendering framework. The key idea is to perform object-centric voxelization to capture the 3D nature of the scene, which infers the probability distribution over objects at individual spatial locations. These voxel features evolve over time through a canonical-space deformation function, forming the basis for global representation learning via slot attention. The voxel features and global features are complementary and are both leveraged by a compositional NeRF decoder for volume rendering. DynaVol remarkably outperforms existing approaches for unsupervised dynamic scene decomposition. Once trained, the explicitly meaningful voxel features enable additional capabilities that 2D scene decomposition methods cannot achieve: it is possible to freely edit the geometric shapes or manipulate the motion trajectories of the objects.
Authors: Moucheng Xu, Yukun Zhou, Chen Jin, Marius de Groot, Daniel C. Alexander, Neil P. Oxtoby, Yipeng Hu, Joseph Jacob
In this paper, we study pseudo-labelling. Pseudo-labelling employs raw inferences on unlabelled data as pseudo-labels for self-training. We elucidate the empirical successes of pseudo-labelling by establishing a link between this technique and the Expectation Maximisation algorithm. Through this, we realise that the original pseudo-labelling serves as an empirical estimation of its more comprehensive underlying formulation. Following this insight, we present a full generalisation of pseudo-labels under Bayes' theorem, termed Bayesian Pseudo Labels. Subsequently, we introduce a variational approach to generate these Bayesian Pseudo Labels, involving the learning of a threshold to automatically select high-quality pseudo labels. In the remainder of the paper, we showcase the applications of pseudo-labelling and its generalised form, Bayesian Pseudo-Labelling, in the semi-supervised segmentation of medical images. Specifically, we focus on: 1) 3D binary segmentation of lung vessels from CT volumes; 2) 2D multi-class segmentation of brain tumours from MRI volumes; 3) 3D binary segmentation of whole brain tumours from MRI volumes; and 4) 3D binary segmentation of prostate from MRI volumes. We further demonstrate that pseudo-labels can enhance the robustness of the learned representations. The code is released in the following GitHub repository: https://github.com/moucheng2017/EMSSL
Authors: Hongxu Jiang, Muhammad Imran, Preethika Muralidharan, Anjali Patel, Jake Pensa, Muxuan Liang, Tarik Benidir, Joseph R. Grajo, Jason P. Joseph, Russell Terry, John Michael DiBianco, Li-Ming Su, Yuyin Zhou, Wayne G. Brisbane, Wei Shao
Micro-ultrasound (micro-US) is a novel 29-MHz ultrasound technique that provides 3-4 times higher resolution than traditional ultrasound, potentially enabling low-cost, accurate diagnosis of prostate cancer. Accurate prostate segmentation is crucial for prostate volume measurement, cancer diagnosis, prostate biopsy, and treatment planning. However, prostate segmentation on micro-US is challenging due to artifacts and indistinct borders between the prostate, bladder, and urethra in the midline. This paper presents MicroSegNet, a multi-scale annotation-guided transformer UNet model designed specifically to tackle these challenges. During the training process, MicroSegNet focuses more on regions that are hard to segment (hard regions), characterized by discrepancies between expert and non-expert annotations. We achieve this by proposing an annotation-guided binary cross entropy (AG-BCE) loss that assigns a larger weight to prediction errors in hard regions and a lower weight to prediction errors in easy regions. The AG-BCE loss was seamlessly integrated into the training process through the utilization of multi-scale deep supervision, enabling MicroSegNet to capture global contextual dependencies and local information at various scales. We trained our model using micro-US images from 55 patients, followed by evaluation on 20 patients. Our MicroSegNet model achieved a Dice coefficient of 0.939 and a Hausdorff distance of 2.02 mm, outperforming several state-of-the-art segmentation methods, as well as three human annotators with different experience levels. Our code is publicly available at https://github.com/mirthAI/MicroSegNet and our dataset is publicly available at https://zenodo.org/records/10475293.
Authors: Josef Bengtson, David Nilsson, Che-Tsung Lin, Marcel Büsching, Fredrik Kahl
We present a generalizable novel view synthesis method which enables modifying the visual appearance of an observed scene so rendered views match a target weather or lighting condition without any scene specific training or access to reference views at the target condition. Our method is based on a pretrained generalizable transformer architecture and is fine-tuned on synthetically generated scenes under different appearance conditions. This allows for rendering novel views in a consistent manner for 3D scenes that were not included in the training set, along with the ability to (i) modify their appearance to match the target condition and (ii) smoothly interpolate between different conditions. Experiments on real and synthetic scenes show that our method is able to generate 3D consistent renderings while making realistic appearance changes, including qualitative and quantitative comparisons. Please refer to our project page for video results: https://ava-nvs.github.io/
Authors: Weiming Zhuang, Lingjuan Lyu
Federated learning (FL) enhances data privacy with collaborative in-situ training on decentralized clients. Nevertheless, FL encounters challenges due to non-independent and identically distributed (non-i.i.d) data, leading to potential performance degradation and hindered convergence. While prior studies predominantly addressed the issue of skewed label distribution, our research addresses a crucial yet frequently overlooked problem known as multi-domain FL. In this scenario, clients' data originate from diverse domains with distinct feature distributions, instead of label distributions. To address the multi-domain problem in FL, we propose a novel method called Federated learning Without normalizations (FedWon). FedWon draws inspiration from the observation that batch normalization (BN) faces challenges in effectively modeling the statistics of multiple domains, while existing normalization techniques possess their own limitations. In order to address these issues, FedWon eliminates the normalization layers in FL and reparameterizes convolution layers with scaled weight standardization. Through extensive experimentation on five datasets and five models, our comprehensive experimental results demonstrate that FedWon surpasses both FedAvg and the current state-of-the-art method (FedBN) across all experimental setups, achieving notable accuracy improvements of more than 10% in certain domains. Furthermore, FedWon is versatile for both cross-silo and cross-device FL, exhibiting robust domain generalization capability, showcasing strong performance even with a batch size as small as 1, thereby catering to resource-constrained devices. Additionally, FedWon can also effectively tackle the challenge of skewed label distribution.
Authors: Ho Man Kwan, Ge Gao, Fan Zhang, Andrew Gower, David Bull
Learning-based video compression is currently a popular research topic, offering the potential to compete with conventional standard video codecs. In this context, Implicit Neural Representations (INRs) have previously been used to represent and compress image and video content, demonstrating relatively high decoding speed compared to other methods. However, existing INR-based methods have failed to deliver rate quality performance comparable with the state of the art in video compression. This is mainly due to the simplicity of the employed network architectures, which limit their representation capability. In this paper, we propose HiNeRV, an INR that combines light weight layers with novel hierarchical positional encodings. We employs depth-wise convolutional, MLP and interpolation layers to build the deep and wide network architecture with high capacity. HiNeRV is also a unified representation encoding videos in both frames and patches at the same time, which offers higher performance and flexibility than existing methods. We further build a video codec based on HiNeRV and a refined pipeline for training, pruning and quantization that can better preserve HiNeRV's performance during lossy model compression. The proposed method has been evaluated on both UVG and MCL-JCV datasets for video compression, demonstrating significant improvement over all existing INRs baselines and competitive performance when compared to learning-based codecs (72.3% overall bit rate saving over HNeRV and 43.4% over DCVC on the UVG dataset, measured in PSNR).
Authors: Haeyong Kang, Jaehong Yoon, DaHyun Kim, Sung Ju Hwang, Chang D Yoo
Neural Implicit Representation (NIR) has recently gained significant attention due to its remarkable ability to encode complex and high-dimensional data into representation space and easily reconstruct it through a trainable mapping function. However, NIR methods assume a one-to-one mapping between the target data and representation models regardless of data relevancy or similarity. This results in poor generalization over multiple complex data and limits their efficiency and scalability. Motivated by continual learning, this work investigates how to accumulate and transfer neural implicit representations for multiple complex video data over sequential encoding sessions. To overcome the limitation of NIR, we propose a novel method, Progressive Fourier Neural Representation (PFNR), that aims to find an adaptive and compact sub-module in Fourier space to encode videos in each training session. This sparsified neural encoding allows the neural network to hold free weights, enabling an improved adaptation for future videos. In addition, when learning a representation for a new video, PFNR transfers the representation of previous videos with frozen weights. This design allows the model to continuously accumulate high-quality neural representations for multiple videos while ensuring lossless decoding that perfectly preserves the learned representations for previous videos. We validate our PFNR method on the UVG8/17 and DAVIS50 video sequence benchmarks and achieve impressive performance gains over strong continual learning baselines. The PFNR code is available at https://github.com/ihaeyong/PFNR.git.
Authors: Jing Lin, Ailing Zeng, Shunlin Lu, Yuanhao Cai, Ruimao Zhang, Haoqian Wang, Lei Zhang
In this paper, we present Motion-X, a large-scale 3D expressive whole-body motion dataset. Existing motion datasets predominantly contain body-only poses, lacking facial expressions, hand gestures, and fine-grained pose descriptions. Moreover, they are primarily collected from limited laboratory scenes with textual descriptions manually labeled, which greatly limits their scalability. To overcome these limitations, we develop a whole-body motion and text annotation pipeline, which can automatically annotate motion from either single- or multi-view videos and provide comprehensive semantic labels for each video and fine-grained whole-body pose descriptions for each frame. This pipeline is of high precision, cost-effective, and scalable for further research. Based on it, we construct Motion-X, which comprises 15.6M precise 3D whole-body pose annotations (i.e., SMPL-X) covering 81.1K motion sequences from massive scenes. Besides, Motion-X provides 15.6M frame-level whole-body pose descriptions and 81.1K sequence-level semantic labels. Comprehensive experiments demonstrate the accuracy of the annotation pipeline and the significant benefit of Motion-X in enhancing expressive, diverse, and natural motion generation, as well as 3D whole-body human mesh recovery.
Authors: Kshama Kodthalu Shivashankara, Deepanshi, Afagh Mehri Shervedani, Gari D. Clifford, Matthew A. Reyna, Reza Sameni
We introduce ECG-Image-Kit, an open-source toolbox for generating synthetic ECG images with realistic artifacts from time-series data, and showcase its application in developing algorithms for data augmentation and ECG image digitization. Synthetic data is generated by producing distortionless ECG images on a standard ECG paper background. Subsequently, various distortions, including handwritten text artifacts, wrinkles, creases, and perspective transformations, are applied to these ECG images. The artifacts and text are synthetically generated, excluding personally identifiable information. The toolbox is used for data augmentation in the 2024 PhysioNet Challenge on Digitization and Classification of ECG Images.
As a case study, we employed ECG-Image-Kit to create an ECG image dataset of 21,801 records from the PhysioNet QT database. A denoising convolutional neural network (DnCNN)-based model was developed and trained on this synthetic dataset and used to convert the synthetically generated images back into time-series data for evaluation. SNR was calculated to assess the quality of image digitization compared to the ground truth ECG time-series. The results show an average signal recovery SNR of 11.17 +/- 9.19 dB, indicating the synthetic ECG image dataset's significance for training deep learning models. For clinical evaluation, we measured the error between the estimated and ground-truth time-series data's RR and QT-intervals. The accuracy of the estimated RR and QT-intervals also suggests that the respective clinical parameters are maintained. These results demonstrate the effectiveness of a deep learning-based pipeline in accurately digitizing paper ECGs and highlight a generative approach to digitization.
Authors: Chen Jiang, Hong Liu, Xuzheng Yu, Qing Wang, Yuan Cheng, Jia Xu, Zhongyi Liu, Qingpei Guo, Wei Chu, Ming Yang, Yuan Qi
In recent years, the explosion of web videos makes text-video retrieval increasingly essential and popular for video filtering, recommendation, and search. Text-video retrieval aims to rank relevant text/video higher than irrelevant ones. The core of this task is to precisely measure the cross-modal similarity between texts and videos. Recently, contrastive learning methods have shown promising results for text-video retrieval, most of which focus on the construction of positive and negative pairs to learn text and video representations. Nevertheless, they do not pay enough attention to hard negative pairs and lack the ability to model different levels of semantic similarity. To address these two issues, this paper improves contrastive learning using two novel techniques. First, to exploit hard examples for robust discriminative power, we propose a novel Dual-Modal Attention-Enhanced Module (DMAE) to mine hard negative pairs from textual and visual clues. By further introducing a Negative-aware InfoNCE (NegNCE) loss, we are able to adaptively identify all these hard negatives and explicitly highlight their impacts in the training loss. Second, our work argues that triplet samples can better model fine-grained semantic similarity compared to pairwise samples. We thereby present a new Triplet Partial Margin Contrastive Learning (TPM-CL) module to construct partial order triplet samples by automatically generating fine-grained hard negatives for matched text-video pairs. The proposed TPM-CL designs an adaptive token masking strategy with cross-modal interaction to model subtle semantic differences. Extensive experiments demonstrate that the proposed approach outperforms existing methods on four widely-used text-video retrieval datasets, including MSR-VTT, MSVD, DiDeMo and ActivityNet.
Authors: Luana Barros, Levy Chaves, Sandra Avila
Melanoma is the most severe type of skin cancer due to its ability to cause metastasis. It is more common in black people, often affecting acral regions: palms, soles, and nails. Deep neural networks have shown tremendous potential for improving clinical care and skin cancer diagnosis. Nevertheless, prevailing studies predominantly rely on datasets of white skin tones, neglecting to report diagnostic outcomes for diverse patient skin tones. In this work, we evaluate supervised and self-supervised models in skin lesion images extracted from acral regions commonly observed in black individuals. Also, we carefully curate a dataset containing skin lesions in acral regions and assess the datasets concerning the Fitzpatrick scale to verify performance on black skin. Our results expose the poor generalizability of these models, revealing their favorable performance for lesions on white skin. Neglecting to create diverse datasets, which necessitates the development of specialized models, is unacceptable. Deep neural networks have great potential to improve diagnosis, particularly for populations with limited access to dermatology. However, including black skin lesions is necessary to ensure these populations can access the benefits of inclusive technology.
Authors: Ruiyuan Gao, Kai Chen, Enze Xie, Lanqing Hong, Zhenguo Li, Dit-Yan Yeung, Qiang Xu
Recent advancements in diffusion models have significantly enhanced the data synthesis with 2D control. Yet, precise 3D control in street view generation, crucial for 3D perception tasks, remains elusive. Specifically, utilizing Bird's-Eye View (BEV) as the primary condition often leads to challenges in geometry control (e.g., height), affecting the representation of object shapes, occlusion patterns, and road surface elevations, all of which are essential to perception data synthesis, especially for 3D object detection tasks. In this paper, we introduce MagicDrive, a novel street view generation framework offering diverse 3D geometry controls, including camera poses, road maps, and 3D bounding boxes, together with textual descriptions, achieved through tailored encoding strategies. Besides, our design incorporates a cross-view attention module, ensuring consistency across multiple camera views. With MagicDrive, we achieve high-fidelity street-view synthesis that captures nuanced 3D geometry and various scene descriptions, enhancing tasks like BEV segmentation and 3D object detection.
Authors: Yi-Lin Sung, Jaehong Yoon, Mohit Bansal
Large Vision-Language Models (LVLMs) can understand the world comprehensively by integrating rich information from different modalities, achieving remarkable advancements on various multimodal downstream tasks. However, deploying LVLMs is often problematic due to their massive computational/energy costs and carbon consumption. Such issues make it infeasible to adopt conventional iterative global pruning, which is costly due to computing the Hessian matrix of the entire large model for sparsification. Alternatively, several studies have recently proposed layer-wise pruning approaches to avoid the expensive computation of global pruning and efficiently compress model weights according to their importance within a layer. However, they often suffer from suboptimal model compression due to their lack of a global perspective. To address this limitation in recent efficient pruning methods for large models, we propose Efficient Coarse-to-Fine LayerWise Pruning (ECoFLaP), a two-stage coarse-to-fine weight pruning approach for LVLMs. We first determine the sparsity ratios of different layers or blocks by leveraging the global importance score, which is efficiently computed based on the zeroth-order approximation of the global model gradients. Then, the model performs local layer-wise unstructured weight pruning based on globally-informed sparsity ratios. We validate our proposed method across various multimodal and unimodal models and datasets, demonstrating significant performance improvements over prevalent pruning techniques in the high-sparsity regime.
Authors: Sophia Sanborn, Nina Miolane
We introduce a general method for achieving robust group-invariance in group-equivariant convolutional neural networks ($G$-CNNs), which we call the $G$-triple-correlation ($G$-TC) layer. The approach leverages the theory of the triple-correlation on groups, which is the unique, lowest-degree polynomial invariant map that is also complete. Many commonly used invariant maps--such as the max--are incomplete: they remove both group and signal structure. A complete invariant, by contrast, removes only the variation due to the actions of the group, while preserving all information about the structure of the signal. The completeness of the triple correlation endows the $G$-TC layer with strong robustness, which can be observed in its resistance to invariance-based adversarial attacks. In addition, we observe that it yields measurable improvements in classification accuracy over standard Max $G$-Pooling in $G$-CNN architectures. We provide a general and efficient implementation of the method for any discretized group, which requires only a table defining the group's product structure. We demonstrate the benefits of this method for $G$-CNNs defined on both commutative and non-commutative groups--$SO(2)$, $O(2)$, $SO(3)$, and $O(3)$ (discretized as the cyclic $C8$, dihedral $D16$, chiral octahedral $O$ and full octahedral $O_h$ groups)--acting on $\mathbb{R}^2$ and $\mathbb{R}^3$ on both $G$-MNIST and $G$-ModelNet10 datasets.
Authors: Ali Hatamizadeh, Michael Ranzinger, Shiyi Lan, Jose M. Alvarez, Sanja Fidler, Jan Kautz
Vision Transformers (ViTs) have attracted a lot of popularity in recent years, due to their exceptional capabilities in modeling long-range spatial dependencies and scalability for large scale training. Although the training parallelism of self-attention mechanism plays an important role in retaining great performance, its quadratic complexity baffles the application of ViTs in many scenarios which demand fast inference. This effect is even more pronounced in applications in which autoregressive modeling of input features is required. In Natural Language Processing (NLP), a new stream of efforts has proposed parallelizable models with recurrent formulation that allows for efficient inference in generative applications. Inspired by this trend, we propose a new class of computer vision models, dubbed Vision Retention Networks (ViR), with dual parallel and recurrent formulations, which strike an optimal balance between fast inference and parallel training with competitive performance. In particular, ViR scales favorably for image throughput and memory consumption in tasks that require higher-resolution images due to its flexible formulation in processing large sequence lengths. The ViR is the first attempt to realize dual parallel and recurrent equivalency in a general vision backbone for recognition tasks. We have validated the effectiveness of ViR through extensive experiments with different dataset sizes and various image resolutions and achieved competitive performance. Code: https://github.com/NVlabs/ViR
Authors: Keqing Du, Xinyu Yang, Hang Chen
Integrating deep learning and causal discovery has increased the interpretability of Temporal Action Segmentation (TAS) tasks. However, frame-level causal relationships exist many complicated noises outside the segment-level, making it infeasible to directly express macro action semantics. Thus, we propose Causal Abstraction Segmentation Refiner (CASR), which can refine TAS results from various models by enhancing video causality in marginalizing frame-level casual relationships. Specifically, we define the equivalent frame-level casual model and segment-level causal model, so that the causal adjacency matrix constructed from marginalized frame-level causal relationships has the ability to represent the segmnet-level causal relationships. CASR works out by reducing the difference in the causal adjacency matrix between we constructed and pre-segmentation results of backbone models. In addition, we propose a novel evaluation metric Causal Edit Distance (CED) to evaluate the causal interpretability. Extensive experimental results on mainstream datasets indicate that CASR significantly surpasses existing various methods in action segmentation performance, as well as in causal explainability and generalization.
Authors: Jia Huang, Peng Jiang, Alvika Gautam, Srikanth Saripalli
Predicting pedestrian behavior is the key to ensure safety and reliability of autonomous vehicles. While deep learning methods have been promising by learning from annotated video frame sequences, they often fail to fully grasp the dynamic interactions between pedestrians and traffic, crucial for accurate predictions. These models also lack nuanced common sense reasoning. Moreover, the manual annotation of datasets for these models is expensive and challenging to adapt to new situations. The advent of Vision Language Models (VLMs) introduces promising alternatives to these issues, thanks to their advanced visual and causal reasoning skills. To our knowledge, this research is the first to conduct both quantitative and qualitative evaluations of VLMs in the context of pedestrian behavior prediction for autonomous driving. We evaluate GPT-4V(ision) on publicly available pedestrian datasets: JAAD and WiDEVIEW. Our quantitative analysis focuses on GPT-4V's ability to predict pedestrian behavior in current and future frames. The model achieves a 57% accuracy in a zero-shot manner, which, while impressive, is still behind the state-of-the-art domain-specific models (70%) in predicting pedestrian crossing actions. Qualitatively, GPT-4V shows an impressive ability to process and interpret complex traffic scenarios, differentiate between various pedestrian behaviors, and detect and analyze groups. However, it faces challenges, such as difficulty in detecting smaller pedestrians and assessing the relative motion between pedestrians and the ego vehicle.
Authors: Weijie Li, Yang Wei, Tianpeng Liu, Yuenan Hou, Yongxiang Liu, Li Liu
Recently, the emergence of a large number of Synthetic Aperture Radar (SAR) sensors and target datasets has made it possible to unify downstream tasks with self-supervised learning techniques, which can pave the way for building the foundation model in the SAR target recognition field. The major challenge of self-supervised learning for SAR target recognition lies in the generalizable representation learning in low data quality and noise.To address the aforementioned problem, we propose a knowledge-guided predictive architecture that uses local masked patches to predict the multiscale SAR feature representations of unseen context. The core of the proposed architecture lies in combining traditional SAR domain feature extraction with state-of-the-art scalable self-supervised learning for accurate generalized feature representations. The proposed framework is validated on various downstream datasets (MSTAR, FUSAR-Ship, SAR-ACD and SSDD), and can bring consistent performance improvement for SAR target recognition. The experimental results strongly demonstrate the unified performance improvement of the self-supervised learning technique for SAR target recognition across diverse targets, scenes and sensors.
Authors: Shuanghao Bai, Min Zhang, Wanqi Zhou, Siteng Huang, Zhirong Luan, Donglin Wang, Badong Chen
Recently, despite the unprecedented success of large pre-trained visual-language models (VLMs) on a wide range of downstream tasks, the real-world unsupervised domain adaptation (UDA) problem is still not well explored. Therefore, in this paper, we first experimentally demonstrate that the unsupervised-trained VLMs can significantly reduce the distribution discrepancy between source and target domains, thereby improving the performance of UDA. However, a major challenge for directly deploying such models on downstream UDA tasks is prompt engineering, which requires aligning the domain knowledge of source and target domains, since the performance of UDA is severely influenced by a good domain-invariant representation. We further propose a Prompt-based Distribution Alignment (PDA) method to incorporate the domain knowledge into prompt learning. Specifically, PDA employs a two-branch prompt-tuning paradigm, namely base branch and alignment branch. The base branch focuses on integrating class-related representation into prompts, ensuring discrimination among different classes. To further minimize domain discrepancy, for the alignment branch, we construct feature banks for both the source and target domains and propose image-guided feature tuning (IFT) to make the input attend to feature banks, which effectively integrates self-enhanced and cross-domain features into the model. In this way, these two branches can be mutually promoted to enhance the adaptation of VLMs for UDA. We conduct extensive experiments on three benchmarks to demonstrate that our proposed PDA achieves state-of-the-art performance. The code is available at https://github.com/BaiShuanghao/Prompt-based-Distribution-Alignment.
Authors: Yechi Ma, Neehar Peri, Shuoquan Wei, Wei Hua, Deva Ramanan, Yanan Li, Shu Kong
Autonomous vehicles (AVs) must accurately detect objects from both common and rare classes for safe navigation, motivating the problem of Long-Tailed 3D Object Detection (LT3D). Contemporary LiDAR-based 3D detectors perform poorly on rare classes (e.g., CenterPoint only achieves 5.1 AP on stroller) as it is difficult to recognize objects from sparse LiDAR points alone. RGB images provide visual evidence to help resolve such ambiguities, motivating the study of RGB-LiDAR fusion. In this paper, we delve into a simple late-fusion framework that ensembles independently trained RGB and LiDAR detectors. Unlike recent end-to-end methods which require paired multi-modal training data, our late-fusion approach can easily leverage large-scale uni-modal datasets, significantly improving rare class detection. In particular, we examine three critical components in this late-fusion framework from first principles, including whether to train 2D or 3D RGB detectors, whether to match RGB and LiDAR detections in 3D or the projected 2D image plane, and how to fuse matched detections.Extensive experiments reveal that 2D RGB detectors achieve better recognition accuracy than 3D RGB detectors, matching on the 2D image plane mitigates depth estimation errors, and fusing scores probabilistically with calibration leads to state-of-the-art LT3D performance. Our late-fusion approach achieves 51.4 mAP on the established nuScenes LT3D benchmark, improving over prior work by 5.9 mAP.
Authors: Aditya Murali, Deepak Alapatt, Pietro Mascagni, Armine Vardazaryan, Alain Garcia, Nariaki Okamoto, Guido Costamagna, Didier Mutter, Jacques Marescaux, Bernard Dallemagne, Nicolas Padoy
This technical report provides a detailed overview of Endoscapes, a dataset of laparoscopic cholecystectomy (LC) videos with highly intricate annotations targeted at automated assessment of the Critical View of Safety (CVS). Endoscapes comprises 201 LC videos with frames annotated sparsely but regularly with segmentation masks, bounding boxes, and CVS assessment by three different clinical experts. Altogether, there are 11090 frames annotated with CVS and 1933 frames annotated with tool and anatomy bounding boxes from the 201 videos, as well as an additional 422 frames from 50 of the 201 videos annotated with tool and anatomy segmentation masks. In this report, we provide detailed dataset statistics (size, class distribution, dataset splits, etc.) and a comprehensive performance benchmark for instance segmentation, object detection, and CVS prediction. The dataset and model checkpoints are publically available at https://github.com/CAMMA-public/Endoscapes.
Authors: Hualie Jiang, Rui Xu, Minglang Tan, Wenjie Jiang
Omnidirectional stereo matching (OSM) is an essential and reliable means for $360^{\circ}$ depth sensing. However, following earlier works on conventional stereo matching, prior state-of-the-art (SOTA) methods rely on a 3D encoder-decoder block to regularize the cost volume, causing the whole system complicated and sub-optimal results. Recently, the Recurrent All-pairs Field Transforms (RAFT) based approach employs the recurrent update in 2D and has efficiently improved image-matching tasks, ie, optical flow, and stereo matching. To bridge the gap between OSM and RAFT, we mainly propose an opposite adaptive weighting scheme to seamlessly transform the outputs of spherical sweeping of OSM into the required inputs for the recurrent update, thus creating a recurrent omnidirectional stereo matching (RomniStereo) algorithm. Furthermore, we introduce two techniques, ie, grid embedding and adaptive context feature generation, which also contribute to RomniStereo's performance. Our best model improves the average MAE metric by 40.7\% over the previous SOTA baseline across five datasets. When visualizing the results, our models demonstrate clear advantages on both synthetic and realistic examples. The code is available at \url{https://github.com/HalleyJiang/RomniStereo}.
Authors: Mohammad Khateri, Morteza Ghahremani, Alejandra Sierra, Jussi Tohka
The inability to acquire clean high-resolution (HR) electron microscopy (EM) images over a large brain tissue volume hampers many neuroscience studies. To address this challenge, we propose a deep-learning-based image super-resolution (SR) approach to computationally reconstruct clean HR 3D-EM with a large field of view (FoV) from noisy low-resolution (LR) acquisition. Our contributions are I) Investigating training with no-clean references for $\ell_2$ and $\ell_1$ loss functions; II) Introducing a novel network architecture, named EMSR, for enhancing the resolution of LR EM images while reducing inherent noise; and, III) Comparing different training strategies including using acquired LR and HR image pairs, i.e., real pairs with no-clean references contaminated with real corruptions, the pairs of synthetic LR and acquired HR, as well as acquired LR and denoised HR pairs. Experiments with nine brain datasets showed that training with real pairs can produce high-quality super-resolved results, demonstrating the feasibility of training with non-clean references for both loss functions. Additionally, comparable results were observed, both visually and numerically, when employing denoised and noisy references for training. Moreover, utilizing the network trained with synthetically generated LR images from HR counterparts proved effective in yielding satisfactory SR results, even in certain cases, outperforming training with real pairs. The proposed SR network was compared quantitatively and qualitatively with several established SR techniques, showcasing either the superiority or competitiveness of the proposed method in mitigating noise while recovering fine details.
Authors: Sadeep Jayasumana, Srikumar Ramalingam, Andreas Veit, Daniel Glasner, Ayan Chakrabarti, Sanjiv Kumar
As with many machine learning problems, the progress of image generation methods hinges on good evaluation metrics. One of the most popular is the Frechet Inception Distance (FID). FID estimates the distance between a distribution of Inception-v3 features of real images, and those of images generated by the algorithm. We highlight important drawbacks of FID: Inception's poor representation of the rich and varied content generated by modern text-to-image models, incorrect normality assumptions, and poor sample complexity. We call for a reevaluation of FID's use as the primary quality metric for generated images. We empirically demonstrate that FID contradicts human raters, it does not reflect gradual improvement of iterative text-to-image models, it does not capture distortion levels, and that it produces inconsistent results when varying the sample size. We also propose an alternative new metric, CMMD, based on richer CLIP embeddings and the maximum mean discrepancy distance with the Gaussian RBF kernel. It is an unbiased estimator that does not make any assumptions on the probability distribution of the embeddings and is sample efficient. Through extensive experiments and analysis, we demonstrate that FID-based evaluations of text-to-image models may be unreliable, and that CMMD offers a more robust and reliable assessment of image quality.
Authors: Georgios Ioannides, Aman Chadha, Aaron Elkins
We propose the Multi-Head Gaussian Adaptive Attention Mechanism (GAAM), a novel probabilistic attention framework, and the Gaussian Adaptive Transformer (GAT), designed to enhance information aggregation across multiple modalities, including Speech, Text and Vision. GAAM integrates learnable mean and variance into its attention mechanism, implemented in a Multi-Headed framework enabling it to collectively model any Probability Distribution for dynamic recalibration of feature significance. This method demonstrates significant improvements, especially with highly non-stationary data, surpassing the state-of-the-art attention techniques in model performance (up to approximately +20% in accuracy) by identifying key elements within the feature space. GAAM's compatibility with dot-product-based attention models and relatively low number of parameters showcases its adaptability and potential to boost existing attention frameworks. Empirically, GAAM exhibits superior adaptability and efficacy across a diverse range of tasks, including emotion recognition in speech, image classification, and text classification, thereby establishing its robustness and versatility in handling multi-modal data. Furthermore, we introduce the Importance Factor (IF), a new learning-based metric that enhances the explainability of models trained with GAAM-based methods. Overall, GAAM represents an advancement towards development of better performing and more explainable attention models across multiple modalities.
Authors: Nicolás Ayobi, Santiago Rodríguez, Alejandra Pérez, Isabela Hernández, Nicolás Aparicio, Eugénie Dessevres, Sebastián Peña, Jessica Santander, Juan Ignacio Caicedo, Nicolás Fernández, Pablo Arbeláez
This paper presents the Holistic and Multi-Granular Surgical Scene Understanding of Prostatectomies (GraSP) dataset, a curated benchmark that models surgical scene understanding as a hierarchy of complementary tasks with varying levels of granularity. Our approach enables a multi-level comprehension of surgical activities, encompassing long-term tasks such as surgical phases and steps recognition and short-term tasks including surgical instrument segmentation and atomic visual actions detection. To exploit our proposed benchmark, we introduce the Transformers for Actions, Phases, Steps, and Instrument Segmentation (TAPIS) model, a general architecture that combines a global video feature extractor with localized region proposals from an instrument segmentation model to tackle the multi-granularity of our benchmark. Through extensive experimentation, we demonstrate the impact of including segmentation annotations in short-term recognition tasks, highlight the varying granularity requirements of each task, and establish TAPIS's superiority over previously proposed baselines and conventional CNN-based models. Additionally, we validate the robustness of our method across multiple public benchmarks, confirming the reliability and applicability of our dataset. This work represents a significant step forward in Endoscopic Vision, offering a novel and comprehensive framework for future research towards a holistic understanding of surgical procedures.
Authors: Changkun Liu, Yukun Zhao, Tristan Braud
Recent years have seen significant improvement in absolute camera pose estimation, paving the way for pervasive markerless Augmented Reality (AR). However, accurate absolute pose estimation techniques are computation- and storage-heavy, requiring computation offloading. As such, AR systems rely on visual-inertial odometry (VIO) to track the device's relative pose between requests to the server. However, VIO suffers from drift, requiring frequent absolute repositioning. This paper introduces MobileARLoc, a new framework for on-device large-scale markerless mobile AR that combines an absolute pose regressor (APR) with a local VIO tracking system. Absolute pose regressors (APRs) provide fast on-device pose estimation at the cost of reduced accuracy. To address APR accuracy and reduce VIO drift, MobileARLoc creates a feedback loop where VIO pose estimations refine the APR predictions. The VIO system identifies reliable predictions of APR, which are then used to compensate for the VIO drift. We comprehensively evaluate MobileARLoc through dataset simulations. MobileARLoc halves the error compared to the underlying APR and achieve fast (80\,ms) on-device inference speed.
Authors: Lincan Li, Wei Shao, Wei Dong, Yijun Tian, Qiming Zhang, Kaixiang Yang, Wenjie Zhang
The aspiration of the next generation's autonomous driving (AD) technology relies on the dedicated integration and interaction among intelligent perception, prediction, planning, and low-level control. There has been a huge bottleneck regarding the upper bound of autonomous driving algorithm performance, a consensus from academia and industry believes that the key to surmount the bottleneck lies in data-centric autonomous driving technology. Recent advancement in AD simulation, closed-loop model training, and AD big data engine have gained some valuable experience. However, there is a lack of systematic knowledge and deep understanding regarding how to build efficient data-centric AD technology for AD algorithm self-evolution and better AD big data accumulation. To fill in the identified research gaps, this article will closely focus on reviewing the state-of-the-art data-driven autonomous driving technologies, with an emphasis on the comprehensive taxonomy of autonomous driving datasets characterized by milestone generations, key features, data acquisition settings, etc. Furthermore, we provide a systematic review of the existing benchmark closed-loop AD big data pipelines from the industrial frontier, including the procedure of closed-loop frameworks, key technologies, and empirical studies. Finally, the future directions, potential applications, limitations and concerns are discussed to arouse efforts from both academia and industry for promoting the further development of autonomous driving. The project repository is available at: https://github.com/LincanLi98/Awesome-Data-Centric-Autonomous-Driving.
Authors: Ismail Nejjar, Gaetan Frusque, Florent Forest, Olga Fink
Unsupervised Domain Adaptation for Regression (UDAR) aims to adapt a model from a labeled source domain to an unlabeled target domain for regression tasks. Recent successful works in UDAR mostly focus on subspace alignment, involving the alignment of a selected subspace within the entire feature space. This contrasts with the feature alignment methods used for classification, which aim at aligning the entire feature space and have proven effective but are less so in regression settings. Specifically, while classification aims to identify separate clusters across the entire embedding dimension, regression induces less structure in the data representation, necessitating additional guidance for efficient alignment. In this paper, we propose an effective method for UDAR by incorporating guidance from uncertainty. Our approach serves a dual purpose: providing a measure of confidence in predictions and acting as a regularization of the embedding space. Specifically, we leverage the Deep Evidential Learning framework, which outputs both predictions and uncertainties for each input sample. We propose aligning the parameters of higher-order evidential distributions between the source and target domains using traditional alignment methods at the feature or posterior level. Additionally, we propose to augment the feature space representation by mixing source samples with pseudo-labeled target samples based on label similarity. This cross-domain mixing strategy produces more realistic samples than random mixing and introduces higher uncertainty, facilitating further alignment. We demonstrate the effectiveness of our approach on four benchmarks for UDAR, on which we outperform existing methods.
Authors: Rameshwar Mishra, A V Subramanyam
Advancements in generative models have sparked significant interest in generating images while adhering to specific structural guidelines. Scene graph to image generation is one such task of generating images which are consistent with the given scene graph. However, the complexity of visual scenes poses a challenge in accurately aligning objects based on specified relations within the scene graph. Existing methods approach this task by first predicting a scene layout and generating images from these layouts using adversarial training. In this work, we introduce a novel approach to generate images from scene graphs which eliminates the need of predicting intermediate layouts. We leverage pre-trained text-to-image diffusion models and CLIP guidance to translate graph knowledge into images. Towards this, we first pre-train our graph encoder to align graph features with CLIP features of corresponding images using a GAN based training. Further, we fuse the graph features with CLIP embedding of object labels present in the given scene graph to create a graph consistent CLIP guided conditioning signal. In the conditioning input, object embeddings provide coarse structure of the image and graph features provide structural alignment based on relationships among objects. Finally, we fine tune a pre-trained diffusion model with the graph consistent conditioning signal with reconstruction and CLIP alignment loss. Elaborate experiments reveal that our method outperforms existing methods on standard benchmarks of COCO-stuff and Visual Genome dataset.