Rama Chellappa

Micah Goldblum, Hossein Souri, Renkun Ni et al. · gatech

Neural network based computer vision systems are typically built on a backbone, a pretrained or randomly initialized feature extractor. Several years ago, the default option was an ImageNet-trained convolutional neural network. However, the recent past has seen the emergence of countless backbones pretrained using various algorithms and datasets. While this abundance of choice has led to performance increases for a range of systems, it is difficult for practitioners to make informed decisions about which backbone to choose. Battle of the Backbones (BoB) makes this choice easier by benchmarking a diverse suite of pretrained models, including vision-language models, those trained via self-supervised learning, and the Stable Diffusion backbone, across a diverse set of computer vision tasks ranging from classification to object detection to OOD generalization and more. Furthermore, BoB sheds light on promising directions for the research community to advance computer vision by illuminating strengths and weakness of existing approaches through a comprehensive analysis conducted on more than 1500 training runs. While vision transformers (ViTs) and self-supervised learning (SSL) are increasingly popular, we find that convolutional neural networks pretrained in a supervised fashion on large training sets still perform best on most tasks among the models we consider. Moreover, in apples-to-apples comparisons on the same architectures and similarly sized pretraining datasets, we find that SSL backbones are highly competitive, indicating that future works should perform SSL pretraining with advanced architectures and larger pretraining datasets. We release the raw results of our experiments along with code that allows researchers to put their own backbones through the gauntlet here: https://github.com/hsouri/Battle-of-the-Backbones

Qiao Gu, Alihusein Kuwajerwala, Sacha Morin et al. · mila, mit

For robots to perform a wide variety of tasks, they require a 3D representation of the world that is semantically rich, yet compact and efficient for task-driven perception and planning. Recent approaches have attempted to leverage features from large vision-language models to encode semantics in 3D representations. However, these approaches tend to produce maps with per-point feature vectors, which do not scale well in larger environments, nor do they contain semantic spatial relationships between entities in the environment, which are useful for downstream planning. In this work, we propose ConceptGraphs, an open-vocabulary graph-structured representation for 3D scenes. ConceptGraphs is built by leveraging 2D foundation models and fusing their output to 3D by multi-view association. The resulting representations generalize to novel semantic classes, without the need to collect large 3D datasets or finetune models. We demonstrate the utility of this representation through a number of downstream planning tasks that are specified through abstract (language) prompts and require complex reasoning over spatial and semantic concepts. (Project page: https://concept-graphs.github.io/ Explainer video: https://youtu.be/mRhNkQwRYnc )

Cheng Peng, Pengfei Guo, S. Kevin Zhou et al.

Magnetic Resonance (MR) image reconstruction from under-sampled acquisition promises faster scanning time. To this end, current State-of-The-Art (SoTA) approaches leverage deep neural networks and supervised training to learn a recovery model. While these approaches achieve impressive performances, the learned model can be fragile on unseen degradation, e.g. when given a different acceleration factor. These methods are also generally deterministic and provide a single solution to an ill-posed problem; as such, it can be difficult for practitioners to understand the reliability of the reconstruction. We introduce DiffuseRecon, a novel diffusion model-based MR reconstruction method. DiffuseRecon guides the generation process based on the observed signals and a pre-trained diffusion model, and does not require additional training on specific acceleration factors. DiffuseRecon is stochastic in nature and generates results from a distribution of fully-sampled MR images; as such, it allows us to explicitly visualize different potential reconstruction solutions. Lastly, DiffuseRecon proposes an accelerated, coarse-to-fine Monte-Carlo sampling scheme to approximate the most likely reconstruction candidate. The proposed DiffuseRecon achieves SoTA performances reconstructing from raw acquisition signals in fastMRI and SKM-TEA. Code will be open-sourced at www.github.com/cpeng93/DiffuseRecon.

Milind Naphade, Shuo Wang, David C. Anastasiu et al. · mit

The AI City Challenge's seventh edition emphasizes two domains at the intersection of computer vision and artificial intelligence - retail business and Intelligent Traffic Systems (ITS) - that have considerable untapped potential. The 2023 challenge had five tracks, which drew a record-breaking number of participation requests from 508 teams across 46 countries. Track 1 was a brand new track that focused on multi-target multi-camera (MTMC) people tracking, where teams trained and evaluated using both real and highly realistic synthetic data. Track 2 centered around natural-language-based vehicle track retrieval. Track 3 required teams to classify driver actions in naturalistic driving analysis. Track 4 aimed to develop an automated checkout system for retail stores using a single view camera. Track 5, another new addition, tasked teams with detecting violations of the helmet rule for motorcyclists. Two leader boards were released for submissions based on different methods: a public leader board for the contest where external private data wasn't allowed and a general leader board for all results submitted. The participating teams' top performances established strong baselines and even outperformed the state-of-the-art in the proposed challenge tracks.

Shraman Pramanick, Li Jing, Sayan Nag et al. · openai

Vision-language pre-training (VLP) has recently proven highly effective for various uni- and multi-modal downstream applications. However, most existing end-to-end VLP methods use high-resolution image-text box data to perform well on fine-grained region-level tasks, such as object detection, segmentation, and referring expression comprehension. Unfortunately, such high-resolution images with accurate bounding box annotations are expensive to collect and use for supervision at scale. In this work, we propose VoLTA (Vision-Language Transformer with weakly-supervised local-feature Alignment), a new VLP paradigm that only utilizes image-caption data but achieves fine-grained region-level image understanding, eliminating the use of expensive box annotations. VoLTA adopts graph optimal transport-based weakly-supervised alignment on local image patches and text tokens to germinate an explicit, self-normalized, and interpretable low-level matching criterion. In addition, VoLTA pushes multi-modal fusion deep into the uni-modal backbones during pre-training and removes fusion-specific transformer layers, further reducing memory requirements. Extensive experiments on a wide range of vision- and vision-language downstream tasks demonstrate the effectiveness of VoLTA on fine-grained applications without compromising the coarse-grained downstream performance, often outperforming methods using significantly more caption and box annotations.

Shraman Pramanick, Yale Song, Sayan Nag et al. · microsoft-research, uw

Video-language pre-training (VLP) has become increasingly important due to its ability to generalize to various vision and language tasks. However, existing egocentric VLP frameworks utilize separate video and language encoders and learn task-specific cross-modal information only during fine-tuning, limiting the development of a unified system. In this work, we introduce the second generation of egocentric video-language pre-training (EgoVLPv2), a significant improvement from the previous generation, by incorporating cross-modal fusion directly into the video and language backbones. EgoVLPv2 learns strong video-text representation during pre-training and reuses the cross-modal attention modules to support different downstream tasks in a flexible and efficient manner, reducing fine-tuning costs. Moreover, our proposed fusion in the backbone strategy is more lightweight and compute-efficient than stacking additional fusion-specific layers. Extensive experiments on a wide range of VL tasks demonstrate the effectiveness of EgoVLPv2 by achieving consistent state-of-the-art performance over strong baselines across all downstream. Our project page can be found at https://shramanpramanick.github.io/EgoVLPv2/.

Anshul Shah, Aniket Roy, Ketul Shah et al.

Supervised learning of skeleton sequence encoders for action recognition has received significant attention in recent times. However, learning such encoders without labels continues to be a challenging problem. While prior works have shown promising results by applying contrastive learning to pose sequences, the quality of the learned representations is often observed to be closely tied to data augmentations that are used to craft the positives. However, augmenting pose sequences is a difficult task as the geometric constraints among the skeleton joints need to be enforced to make the augmentations realistic for that action. In this work, we propose a new contrastive learning approach to train models for skeleton-based action recognition without labels. Our key contribution is a simple module, HaLP - to Hallucinate Latent Positives for contrastive learning. Specifically, HaLP explores the latent space of poses in suitable directions to generate new positives. To this end, we present a novel optimization formulation to solve for the synthetic positives with an explicit control on their hardness. We propose approximations to the objective, making them solvable in closed form with minimal overhead. We show via experiments that using these generated positives within a standard contrastive learning framework leads to consistent improvements across benchmarks such as NTU-60, NTU-120, and PKU-II on tasks like linear evaluation, transfer learning, and kNN evaluation. Our code will be made available at https://github.com/anshulbshah/HaLP.

Vivekjyoti Banerjee, Abhay Yadav, Rama Chellappa et al.

3D Gaussian Splatting (3DGS) enables real-time novel view synthesis by representing scenes as collections of anisotropic Gaussians optimized via differentiable rasterization. However, standard pixel-space losses (L1, SSIM) constrain only aggregate reconstruction error, permitting the optimization to redistribute error across frequency scales. This leads to oversmoothing and structural artifacts, particularly in sparse-view settings where supervision is limited. We propose KC-3DGS, which augments 3DGS training with wavelet-domain supervision based on natural image statistics. Our method combines three components: (1) a multi-scale wavelet coefficient alignment loss that explicitly penalizes missing high-frequency detail, (2) a supervised kurtosis concentration loss that encourages rendered images to match the heavy-tailed frequency statistics of ground-truth images, and (3) a cross-band covariance penalty that promotes frequency specialization. We provide theoretical analysis showing that pixel-space losses admit a family of indistinguishable perturbations under wavelet redistribution, and that our joint objective excludes degenerate solutions. Experiments across MipNeRF360, Tanks&Temples, MVImgNet, DeepBlending, and WRIVA-ULTRRA demonstrate consistent improvements in perceptual quality. On the challenging WRIVA-ULTRRA outdoor dataset, KC-3DGS achieves a 9.48% improvement in DreamSim while also improving PSNR, SSIM, and LPIPS. In sparse-view settings with only 12 training images, our method improves PSNR by up to 0.5 dB on MipNeRF360 while maintaining perceptual quality. The approach integrates seamlessly into existing 3DGS pipelines as a plug-and-play regularization strategy.

Milind Naphade, Shuo Wang, David C. Anastasiu et al.

The 6th edition of the AI City Challenge specifically focuses on problems in two domains where there is tremendous unlocked potential at the intersection of computer vision and artificial intelligence: Intelligent Traffic Systems (ITS), and brick and mortar retail businesses. The four challenge tracks of the 2022 AI City Challenge received participation requests from 254 teams across 27 countries. Track 1 addressed city-scale multi-target multi-camera (MTMC) vehicle tracking. Track 2 addressed natural-language-based vehicle track retrieval. Track 3 was a brand new track for naturalistic driving analysis, where the data were captured by several cameras mounted inside the vehicle focusing on driver safety, and the task was to classify driver actions. Track 4 was another new track aiming to achieve retail store automated checkout using only a single view camera. We released two leader boards for submissions based on different methods, including a public leader board for the contest, where no use of external data is allowed, and a general leader board for all submitted results. The top performance of participating teams established strong baselines and even outperformed the state-of-the-art in the proposed challenge tracks.

Shraman Pramanick, Ewa M. Nowara, Joshua Gleason et al.

Predicting the geographic location (geo-localization) from a single ground-level RGB image taken anywhere in the world is a very challenging problem. The challenges include huge diversity of images due to different environmental scenarios, drastic variation in the appearance of the same location depending on the time of the day, weather, season, and more importantly, the prediction is made from a single image possibly having only a few geo-locating cues. For these reasons, most existing works are restricted to specific cities, imagery, or worldwide landmarks. In this work, we focus on developing an efficient solution to planet-scale single-image geo-localization. To this end, we propose TransLocator, a unified dual-branch transformer network that attends to tiny details over the entire image and produces robust feature representation under extreme appearance variations. TransLocator takes an RGB image and its semantic segmentation map as inputs, interacts between its two parallel branches after each transformer layer, and simultaneously performs geo-localization and scene recognition in a multi-task fashion. We evaluate TransLocator on four benchmark datasets - Im2GPS, Im2GPS3k, YFCC4k, YFCC26k and obtain 5.5%, 14.1%, 4.9%, 9.9% continent-level accuracy improvement over the state-of-the-art. TransLocator is also validated on real-world test images and found to be more effective than previous methods.

Sai Saketh Rambhatla, Ishan Misra, Rama Chellappa et al.

We tackle the challenging task of unsupervised object localization in this work. Recently, transformers trained with self-supervised learning have been shown to exhibit object localization properties without being trained for this task. In this work, we present Multiple Object localization with Self-supervised Transformers (MOST) that uses features of transformers trained using self-supervised learning to localize multiple objects in real world images. MOST analyzes the similarity maps of the features using box counting; a fractal analysis tool to identify tokens lying on foreground patches. The identified tokens are then clustered together, and tokens of each cluster are used to generate bounding boxes on foreground regions. Unlike recent state-of-the-art object localization methods, MOST can localize multiple objects per image and outperforms SOTA algorithms on several object localization and discovery benchmarks on PASCAL-VOC 07, 12 and COCO20k datasets. Additionally, we show that MOST can be used for self-supervised pre-training of object detectors, and yields consistent improvements on fully, semi-supervised object detection and unsupervised region proposal generation.

Cheng Peng, Rama Chellappa

We present Progressively Deblurring Radiance Field (PDRF), a novel approach to efficiently reconstruct high quality radiance fields from blurry images. While current State-of-The-Art (SoTA) scene reconstruction methods achieve photo-realistic rendering results from clean source views, their performances suffer when the source views are affected by blur, which is commonly observed for images in the wild. Previous deblurring methods either do not account for 3D geometry, or are computationally intense. To addresses these issues, PDRF, a progressively deblurring scheme in radiance field modeling, accurately models blur by incorporating 3D scene context. PDRF further uses an efficient importance sampling scheme, which results in fast scene optimization. Specifically, PDRF proposes a Coarse Ray Renderer to quickly estimate voxel density and feature; a Fine Voxel Renderer is then used to achieve high quality ray tracing. We perform extensive experiments and show that PDRF is 15X faster than previous SoTA while achieving better performance on both synthetic and real scenes.

Yuxiang Guo, Cheng Peng, Chun Pong Lau et al.

Whole-body-based human authentication is a promising approach for remote biometrics scenarios. Current literature focuses on either body recognition based on RGB images or gait recognition based on body shapes and walking patterns; both have their advantages and drawbacks. In this work, we propose Dual-Modal Ensemble (DME), which combines both RGB and silhouette data to achieve more robust performances for indoor and outdoor whole-body based recognition. Within DME, we propose GaitPattern, which is inspired by the double helical gait pattern used in traditional gait analysis. The GaitPattern contributes to robust identification performance over a large range of viewing angles. Extensive experimental results on the CASIA-B dataset demonstrate that the proposed method outperforms state-of-the-art recognition systems. We also provide experimental results using the newly collected BRIAR dataset.

Chrisopher B. Nalty, Neehar Peri, Joshua Gleason et al.

Person recognition at a distance entails recognizing the identity of an individual appearing in images or videos collected by long-range imaging systems such as drones or surveillance cameras. Despite recent advances in deep convolutional neural networks (DCNNs), this remains challenging. Images or videos collected by long-range cameras often suffer from atmospheric turbulence, blur, low-resolution, unconstrained poses, and poor illumination. In this paper, we provide a brief survey of recent advances in person recognition at a distance. In particular, we review recent work in multi-spectral face verification, person re-identification, and gait-based analysis techniques. Furthermore, we discuss the merits and drawbacks of existing approaches and identify important, yet under explored challenges for deploying remote person recognition systems in-the-wild.

Yuxin Wen, Jonas Geiping, Liam Fowl et al.

Federated learning is particularly susceptible to model poisoning and backdoor attacks because individual users have direct control over the training data and model updates. At the same time, the attack power of an individual user is limited because their updates are quickly drowned out by those of many other users. Existing attacks do not account for future behaviors of other users, and thus require many sequential updates and their effects are quickly erased. We propose an attack that anticipates and accounts for the entire federated learning pipeline, including behaviors of other clients, and ensures that backdoors are effective quickly and persist even after multiple rounds of community updates. We show that this new attack is effective in realistic scenarios where the attacker only contributes to a small fraction of randomly sampled rounds and demonstrate this attack on image classification, next-word prediction, and sentiment analysis.

Arun V. Reddy, Ketul Shah, William Paul et al.

Human action recognition is a challenging problem, particularly when there is high variability in factors such as subject appearance, backgrounds and viewpoint. While deep neural networks (DNNs) have been shown to perform well on action recognition tasks, they typically require large amounts of high-quality labeled data to achieve robust performance across a variety of conditions. Synthetic data has shown promise as a way to avoid the substantial costs and potential ethical concerns associated with collecting and labeling enormous amounts of data in the real-world. However, synthetic data may differ from real data in important ways. This phenomenon, known as \textit{domain shift}, can limit the utility of synthetic data in robotics applications. To mitigate the effects of domain shift, substantial effort is being dedicated to the development of domain adaptation (DA) techniques. Yet, much remains to be understood about how best to develop these techniques. In this paper, we introduce a new dataset called Robot Control Gestures (RoCoG-v2). The dataset is composed of both real and synthetic videos from seven gesture classes, and is intended to support the study of synthetic-to-real domain shift for video-based action recognition. Our work expands upon existing datasets by focusing the action classes on gestures for human-robot teaming, as well as by enabling investigation of domain shift in both ground and aerial views. We present baseline results using state-of-the-art action recognition and domain adaptation algorithms and offer initial insight on tackling the synthetic-to-real and ground-to-air domain shifts.

Yuxiang Guo, Siyuan Huang, Ram Prabhakar et al.

Gait recognition holds the promise of robustly identifying subjects based on walking patterns instead of appearance information. While previous approaches have performed well for curated indoor data, they tend to underperform in unconstrained situations, e.g. in outdoor, long distance scenes, etc. We propose a framework, termed GAit DEtection and Recognition (GADER), for human authentication in challenging outdoor scenarios. Specifically, GADER leverages a Double Helical Signature to detect segments that contain human movement and builds discriminative features through a novel gait recognition method, where only frames containing gait information are used. To further enhance robustness, GADER encodes viewpoint information in its architecture, and distills representation from an auxiliary RGB recognition model, which enables GADER to learn from silhouette and RGB data at training time. At test time, GADER only infers from the silhouette modality. We evaluate our method on multiple State-of-The-Arts(SoTA) gait baselines and demonstrate consistent improvements on indoor and outdoor datasets, especially with a significant 25.2% improvement on unconstrained, remote gait data.

Pirazh Khorramshahi, Vineet Shenoy, Michael Pack et al.

Multi-camera vehicle tracking is one of the most complicated tasks in Computer Vision as it involves distinct tasks including Vehicle Detection, Tracking, and Re-identification. Despite the challenges, multi-camera vehicle tracking has immense potential in transportation applications including speed, volume, origin-destination (O-D), and routing data generation. Several recent works have addressed the multi-camera tracking problem. However, most of the effort has gone towards improving accuracy on high-quality benchmark datasets while disregarding lower camera resolutions, compression artifacts and the overwhelming amount of computational power and time needed to carry out this task on its edge and thus making it prohibitive for large-scale and real-time deployment. Therefore, in this work we shed light on practical issues that should be addressed for the design of a multi-camera tracking system to provide actionable and timely insights. Moreover, we propose a real-time city-scale multi-camera vehicle tracking system that compares favorably to computationally intensive alternatives and handles real-world, low-resolution CCTV instead of idealized and curated video streams. To show its effectiveness, in addition to integration into the Regional Integrated Transportation Information System (RITIS), we participated in the 2021 NVIDIA AI City multi-camera tracking challenge and our method is ranked among the top five performers on the public leaderboard.

Shraman Pramanick, Rama Chellappa, Subhashini Venugopalan

Seeking answers to questions within long scientific research articles is a crucial area of study that aids readers in quickly addressing their inquiries. However, existing question-answering (QA) datasets based on scientific papers are limited in scale and focus solely on textual content. We introduce SPIQA (Scientific Paper Image Question Answering), the first large-scale QA dataset specifically designed to interpret complex figures and tables within the context of scientific research articles across various domains of computer science. Leveraging the breadth of expertise and ability of multimodal large language models (MLLMs) to understand figures, we employ automatic and manual curation to create the dataset. We craft an information-seeking task on interleaved images and text that involves multiple images covering plots, charts, tables, schematic diagrams, and result visualizations. SPIQA comprises 270K questions divided into training, validation, and three different evaluation splits. Through extensive experiments with 12 prominent foundational models, we evaluate the ability of current multimodal systems to comprehend the nuanced aspects of research articles. Additionally, we propose a Chain-of-Thought (CoT) evaluation strategy with in-context retrieval that allows fine-grained, step-by-step assessment and improves model performance. We further explore the upper bounds of performance enhancement with additional textual information, highlighting its promising potential for future research and the dataset's impact on revolutionizing how we interact with scientific literature.

Yuxiang Guo, Faizan Siddiqui, Yang Zhao et al.

Predicting and reasoning how a video would make a human feel is crucial for developing socially intelligent systems. Although Multimodal Large Language Models (MLLMs) have shown impressive video understanding capabilities, they tend to focus more on the semantic content of videos, often overlooking emotional stimuli. Hence, most existing MLLMs fall short in estimating viewers' emotional reactions and providing plausible explanations. To address this issue, we propose StimuVAR, a spatiotemporal Stimuli-aware framework for Video Affective Reasoning (VAR) with MLLMs. StimuVAR incorporates a two-level stimuli-aware mechanism: frame-level awareness and token-level awareness. Frame-level awareness involves sampling video frames with events that are most likely to evoke viewers' emotions. Token-level awareness performs tube selection in the token space to make the MLLM concentrate on emotion-triggered spatiotemporal regions. Furthermore, we create VAR instruction data to perform affective training, steering MLLMs' reasoning strengths towards emotional focus and thereby enhancing their affective reasoning ability. To thoroughly assess the effectiveness of VAR, we provide a comprehensive evaluation protocol with extensive metrics. StimuVAR is the first MLLM-based method for viewer-centered VAR. Experiments demonstrate its superiority in understanding viewers' emotional responses to videos and providing coherent and insightful explanations. Our code is available at https://github.com/EthanG97/StimuVAR

Anshul Shah, Benjamin Lundell, Harpreet Sawhney et al.

We address the problem of extracting key steps from unlabeled procedural videos, motivated by the potential of Augmented Reality (AR) headsets to revolutionize job training and performance. We decompose the problem into two steps: representation learning and key steps extraction. We propose a training objective, Bootstrapped Multi-Cue Contrastive (BMC2) loss to learn discriminative representations for various steps without any labels. Different from prior works, we develop techniques to train a light-weight temporal module which uses off-the-shelf features for self supervision. Our approach can seamlessly leverage information from multiple cues like optical flow, depth or gaze to learn discriminative features for key-steps, making it amenable for AR applications. We finally extract key steps via a tunable algorithm that clusters the representations and samples. We show significant improvements over prior works for the task of key step localization and phase classification. Qualitative results demonstrate that the extracted key steps are meaningful and succinctly represent various steps of the procedural tasks.

Jiang Liu, Chen Wei, Yuxiang Guo et al.

We propose Instruct2Attack (I2A), a language-guided semantic attack that generates semantically meaningful perturbations according to free-form language instructions. We make use of state-of-the-art latent diffusion models, where we adversarially guide the reverse diffusion process to search for an adversarial latent code conditioned on the input image and text instruction. Compared to existing noise-based and semantic attacks, I2A generates more natural and diverse adversarial examples while providing better controllability and interpretability. We further automate the attack process with GPT-4 to generate diverse image-specific text instructions. We show that I2A can successfully break state-of-the-art deep neural networks even under strong adversarial defenses, and demonstrate great transferability among a variety of network architectures.

Pirazh Khorramshahi, Vineet Shenoy, Rama Chellappa

As Computer Vision technologies become more mature for intelligent transportation applications, it is time to ask how efficient and scalable they are for large-scale and real-time deployment. Among these technologies is Vehicle Re-Identification which is one of the key elements in city-scale vehicle analytics systems. Many state-of-the-art solutions for vehicle re-id mostly focus on improving the accuracy on existing re-id benchmarks and often ignore computational complexity. To balance the demands of accuracy and computational efficiency, in this work we propose a simple yet effective hybrid solution empowered by self-supervised training which only uses a single network during inference time and is free of intricate and computation-demanding add-on modules often seen in state-of-the-art approaches. Through extensive experiments, we show our approach, termed Self-Supervised and Boosted VEhicle Re-Identification (SSBVER), is on par with state-of-the-art alternatives in terms of accuracy without introducing any additional overhead during deployment. Additionally we show that our approach, generalizes to different backbone architectures which facilitates various resource constraints and consistently results in a significant accuracy boost.

Guangxing Han, Long Chen, Jiawei Ma et al.

We study multi-modal few-shot object detection (FSOD) in this paper, using both few-shot visual examples and class semantic information for detection, which are complementary to each other by definition. Most of the previous works on multi-modal FSOD are fine-tuning-based which are inefficient for online applications. Moreover, these methods usually require expertise like class names to extract class semantic embedding, which are hard to get for rare classes. Our approach is motivated by the high-level conceptual similarity of (metric-based) meta-learning and prompt-based learning to learn generalizable few-shot and zero-shot object detection models respectively without fine-tuning. Specifically, we combine the few-shot visual classifier and text classifier learned via meta-learning and prompt-based learning respectively to build the multi-modal classifier and detection models. In addition, to fully exploit the pre-trained language models, we propose meta-learning-based cross-modal prompting to generate soft prompts for novel classes present in few-shot visual examples, which are then used to learn the text classifier. Knowledge distillation is introduced to learn the soft prompt generator without using human prior knowledge of class names, which may not be available for rare classes. Our insight is that the few-shot support images naturally include related context information and semantics of the class. We comprehensively evaluate the proposed multi-modal FSOD models on multiple few-shot object detection benchmarks, achieving promising results.

Mahyar Fazlyab, Taha Entesari, Aniket Roy et al.

To improve the robustness of deep classifiers against adversarial perturbations, many approaches have been proposed, such as designing new architectures with better robustness properties (e.g., Lipschitz-capped networks), or modifying the training process itself (e.g., min-max optimization, constrained learning, or regularization). These approaches, however, might not be effective at increasing the margin in the input (feature) space. As a result, there has been an increasing interest in developing training procedures that can directly manipulate the decision boundary in the input space. In this paper, we build upon recent developments in this category by developing a robust training algorithm whose objective is to increase the margin in the output (logit) space while regularizing the Lipschitz constant of the model along vulnerable directions. We show that these two objectives can directly promote larger margins in the input space. To this end, we develop a scalable method for calculating guaranteed differentiable upper bounds on the Lipschitz constant of neural networks accurately and efficiently. The relative accuracy of the bounds prevents excessive regularization and allows for more direct manipulation of the decision boundary. Furthermore, our Lipschitz bounding algorithm exploits the monotonicity and Lipschitz continuity of the activation layers, and the resulting bounds can be used to design new layers with controllable bounds on their Lipschitz constant. Experiments on the MNIST, CIFAR-10, and Tiny-ImageNet data sets verify that our proposed algorithm obtains competitively improved results compared to the state-of-the-art.

Zhusi Zhong, Jie Li, Lulu Bi et al.

Medical image segmentation based on deep learning often fails when deployed on images from a different domain. The domain adaptation methods aim to solve domain-shift challenges, but still face some problems. The transfer learning methods require annotation on the target domain, and the generative unsupervised domain adaptation (UDA) models ignore domain-specific representations, whose generated quality highly restricts segmentation performance. In this study, we propose a novel Structure-Modal Constrained (SMC) UDA framework based on a discriminative paradigm and introduce edge structure as a bridge between domains. The proposed multi-modal learning backbone distills structure information from image texture to distinguish domain-invariant edge structure. With the structure-constrained self-learning and progressive ROI, our methods segment the kidney by locating the 3D spatial structure of the edge. We evaluated SMC-UDA on public renal segmentation datasets, adapting from the labeled source domain (CT) to the unlabeled target domain (CT/MRI). The experiments show that our proposed SMC-UDA has a strong generalization and outperforms generative UDA methods.

Cheng Peng, Haofu Liao, S. Kevin Zhou et al. · amazon-science

It is a long-standing challenge to reconstruct Cone Beam Computed Tomography (CBCT) of the lung under respiratory motion. This work takes a step further to address a challenging setting in reconstructing a multi-phase}4D lung image from just a single}3D CBCT acquisition. To this end, we introduce REpiratory-GAted Synthesis of views, or REGAS. REGAS proposes a self-supervised method to synthesize the undersampled tomographic views and mitigate aliasing artifacts in reconstructed images. This method allows a much better estimation of between-phase Deformation Vector Fields (DVFs), which are used to enhance reconstruction quality from direct observations without synthesis. To address the large memory cost of deep neural networks on high resolution 4D data, REGAS introduces a novel Ray Path Transformation (RPT) that allows for distributed, differentiable forward projections. REGAS require no additional measurements like prior scans, air-flow volume, or breathing velocity. Our extensive experiments show that REGAS significantly outperforms comparable methods in quantitative metrics and visual quality.

Siyuan Huang, Ram Prabhakar Kathirvel, Chun Pong Lau et al.

In this paper, we address the challenging task of whole-body biometric detection, recognition, and identification at distances of up to 500m and large pitch angles of up to 50 degree. We propose an end-to-end system evaluated on diverse datasets, including the challenging Biometric Recognition and Identification at Range (BRIAR) dataset. Our approach involves pre-training the detector on common image datasets and fine-tuning it on BRIAR's complex videos and images. After detection, we extract body images and employ a feature extractor for recognition. We conduct thorough evaluations under various conditions, such as different ranges and angles in indoor, outdoor, and aerial scenarios. Our method achieves an average F1 score of 98.29% at IoU = 0.7 and demonstrates strong performance in recognition accuracy and true acceptance rate at low false acceptance rates compared to existing models. On a test set of 100 subjects with 444 distractors, our model achieves a rank-20 recognition accuracy of 75.13% and a TAR@1%FAR of 54.09%.

Cheng Peng, S. Kevin Zhou, Rama Chellappa

Medical image super-resolution (SR) is an active research area that has many potential applications, including reducing scan time, bettering visual understanding, increasing robustness in downstream tasks, etc. However, applying deep-learning-based SR approaches for clinical applications often encounters issues of domain inconsistency, as the test data may be acquired by different machines or on different organs. In this work, we present a novel algorithm called domain adaptable volumetric super-resolution (DA-VSR) to better bridge the domain inconsistency gap. DA-VSR uses a unified feature extraction backbone and a series of network heads to improve image quality over different planes. Furthermore, DA-VSR leverages the in-plane and through-plane resolution differences on the test data to achieve a self-learned domain adaptation. As such, DA-VSR combines the advantages of a strong feature generator learned through supervised training and the ability to tune to the idiosyncrasies of the test volumes through unsupervised learning. Through experiments, we demonstrate that DA-VSR significantly improves super-resolution quality across numerous datasets of different domains, thereby taking a further step toward real clinical applications.

Guoyizhe Wei, Feng Wang, Anshul Shah et al.

Prompt learning has recently become a very efficient transfer learning paradigm for Contrastive Language Image Pretraining (CLIP) models. Compared with fine-tuning the entire encoder, prompt learning can obtain highly competitive results by optimizing only a small number of parameters, which presents considerably exciting benefits for federated learning applications that prioritizes communication efficiency. However, in this work, we identify that directly transferring prompt learning approaches into federated learning does not yield favorable results since the model often suffers from considerable domain gaps across different clients. To address this issue, we propose ADAPT, a novel domain-aware prompt learning approach that facilitates both intra- and inter-domain prompts across federated participants. The basic idea of ADAPT is that the prompted CLIP should detect the input image's domain correspondence and before making the prediction of its category. Extensive experiments of ADAPT demonstrate its significant efficiency and effectiveness in federated learning. For example, by learning and sharing only 0.08M parameters, our ADAPT attains a 68.4% average accuracy over six domains in the DomainNet dataset, which improves the original CLIP by a large margin of 14.8%.

Gina Wong, Joshua Gleason, Rama Chellappa et al.

Learning models whose predictions are invariant under multiple environments is a promising approach for out-of-distribution generalization. Such models are trained to extract features $X_{\text{inv}}$ where the conditional distribution $Y \mid X_{\text{inv}}$ of the label given the extracted features does not change across environments. Invariant models are also supposed to generalize to shifts in the marginal distribution $p(X_{\text{inv}})$ of the extracted features $X_{\text{inv}}$, a type of shift we call an $\textit{invariant covariate shift}$. However, we show that proposed methods for learning invariant models underperform under invariant covariate shift, either failing to learn invariant models$\unicode{x2014}$even for data generated from simple and well-studied linear-Gaussian models$\unicode{x2014}$or having poor finite-sample performance. To alleviate these problems, we propose $\textit{weighted risk invariance}$ (WRI). Our framework is based on imposing invariance of the loss across environments subject to appropriate reweightings of the training examples. We show that WRI provably learns invariant models, i.e. discards spurious correlations, in linear-Gaussian settings. We propose a practical algorithm to implement WRI by learning the density $p(X_{\text{inv}})$ and the model parameters simultaneously, and we demonstrate empirically that WRI outperforms previous invariant learning methods under invariant covariate shift.

Prakhar Kaushik, Shravan Chaudhari, Ankit Vaidya et al.

We show that deep neural networks trained across diverse tasks exhibit remarkably similar low-dimensional parametric subspaces. We provide the first large-scale empirical evidence that demonstrates that neural networks systematically converge to shared spectral subspaces regardless of initialization, task, or domain. Through mode-wise spectral analysis of over 1100 models - including 500 Mistral-7B LoRAs, 500 Vision Transformers, and 50 LLaMA-8B models - we identify universal subspaces capturing majority variance in just a few principal directions. By applying spectral decomposition techniques to the weight matrices of various architectures trained on a wide range of tasks and datasets, we identify sparse, joint subspaces that are consistently exploited, within shared architectures across diverse tasks and datasets. Our findings offer new insights into the intrinsic organization of information within deep networks and raise important questions about the possibility of discovering these universal subspaces without the need for extensive data and computational resources. Furthermore, this inherent structure has significant implications for model reusability, multi-task learning, model merging, and the development of training and inference-efficient algorithms, potentially reducing the carbon footprint of large-scale neural models.

Deming Li, Abhay Yadav, Cheng Peng et al.

We present SyncFix, a framework that enforces cross-view consistency during the diffusion-based refinement of reconstructed scenes. SyncFix formulates refinement as a joint latent bridge matching problem, synchronizing distorted and clean representations across multiple views to fix the semantic and geometric inconsistencies. This means SyncFix learns a joint conditional over multiple views to enforce consistency throughout the denoising trajectory. Our training is done only on image pairs, but it generalizes naturally to an arbitrary number of views during inference. Moreover, reconstruction quality improves with additional views, with diminishing returns at higher view counts. Qualitative and quantitative results demonstrate that SyncFix consistently generates high-quality reconstructions and surpasses current state-of-the-art baselines, even in the absence of clean reference images. SyncFix achieves even higher fidelity when sparse references are available.

Laura McDaniel, Basudha Pal, Crystal Szczesny et al.

Frailty is a condition in aging medicine characterized by diminished physiological reserve and increased vulnerability to stressors. However, frailty assessment remains subjective, heterogeneous, and difficult to scale in clinical practice. Gait is a sensitive marker of biological aging, capturing multisystem decline before overt disability. Yet the application of modern computer vision to gait-based frailty assessment has been limited by small, imbalanced datasets and a lack of clinically representative benchmarks. In this work, we introduce a publicly available silhouette-based frailty gait dataset collected in a clinically realistic setting, spanning the full frailty spectrum and including older adults who use walking aids. Using this dataset, we evaluate how pretrained gait recognition models can be adapted for frailty classification under limited data conditions. We study both convolutional and hybrid attention-based architectures and show that predictive performance depends primarily on how pretrained representations are transferred rather than architectural complexity alone. Across models, selectively freezing low-level gait representations while allowing higher-level features to adapt yields more stable and generalizable performance than either full fine-tuning or rigid freezing. Conservative handling of class imbalance further improves training stability, and combining complementary learning objectives enhances discrimination between clinically adjacent frailty states. Interpretability analyses reveal consistent model attention to lower-limb and pelvic regions, aligning with established biomechanical correlates of frailty. Together, these findings establish gait-based representation learning as a scalable, non-invasive, and interpretable framework for frailty assessment and support the integration of modern biometric modeling approaches into aging research and clinical practice.

Yuxiang Guo, Anshul Shah, Jiang Liu et al.

Gait recognition holds the promise to robustly identify subjects based on walking patterns instead of appearance information. In recent years, this field has been dominated by learning methods based on two principal input representations: dense silhouette masks or sparse pose keypoints. In this work, we propose a novel, point-based Contour-Pose representation, which compactly expresses both body shape and body parts information. We further propose a local-to-global architecture, called GaitContour, to leverage this novel representation and efficiently compute subject embedding in two stages. The first stage consists of a local transformer that extracts features from five different body regions. The second stage then aggregates the regional features to estimate a global human gait representation. Such a design significantly reduces the complexity of the attention operation and improves efficiency and performance simultaneously. Through large scale experiments, GaitContour is shown to perform significantly better than previous point-based methods, while also being significantly more efficient than silhouette-based methods. On challenging datasets with significant distractors, GaitContour can even outperform silhouette-based methods.

Siyuan Huang, Ram Prabhakar, Yuxiang Guo et al.

Person Re-identification is a research area with significant real world applications. Despite recent progress, existing methods face challenges in robust re-identification in the wild, e.g., by focusing only on a particular modality and on unreliable patterns such as clothing. A generalized method is highly desired, but remains elusive to achieve due to issues such as the trade-off between spatial and temporal resolution and imperfect feature extraction. We propose VILLS (Video-Image Learning to Learn Semantics), a self-supervised method that jointly learns spatial and temporal features from images and videos. VILLS first designs a local semantic extraction module that adaptively extracts semantically consistent and robust spatial features. Then, VILLS designs a unified feature learning and adaptation module to represent image and video modalities in a consistent feature space. By Leveraging self-supervised, large-scale pre-training, VILLS establishes a new State-of-The-Art that significantly outperforms existing image and video-based methods.

Aniket Roy, Maiterya Suin, Anshul Shah et al.

Diffusion models have advanced generative AI significantly in terms of editing and creating naturalistic images. However, efficiently improving generated image quality is still of paramount interest. In this context, we propose a generic "naturalness" preserving loss function, viz., kurtosis concentration (KC) loss, which can be readily applied to any standard diffusion model pipeline to elevate the image quality. Our motivation stems from the projected kurtosis concentration property of natural images, which states that natural images have nearly constant kurtosis values across different band-pass versions of the image. To retain the "naturalness" of the generated images, we enforce reducing the gap between the highest and lowest kurtosis values across the band-pass versions (e.g., Discrete Wavelet Transform (DWT)) of images. Note that our approach does not require any additional guidance like classifier or classifier-free guidance to improve the image quality. We validate the proposed approach for three diverse tasks, viz., (1) personalized few-shot finetuning using text guidance, (2) unconditional image generation, and (3) image super-resolution. Integrating the proposed KC loss has improved the perceptual quality across all these tasks in terms of both FID, MUSIQ score, and user evaluation.

Michael Sprintson, Rama Chellappa, Cheng Peng

We introduce FusionRF, a novel framework for digital surface reconstruction from satellite multispectral and panchromatic images. Current work has demonstrated the increased accuracy of neural photogrammetry for surface reconstruction from optical satellite images compared to algorithmic methods. Common satellites produce both a panchromatic and multispectral image, which contain high spatial and spectral information respectively. Current neural reconstruction methods require multispectral images to be upsampled with a pansharpening method using the spatial data in the panchromatic image. However, these methods may introduce biases and hallucinations due to domain gaps. FusionRF introduces joint image fusion during optimization through a novel cross-resolution kernel that learns to resolve spatial resolution loss present in multispectral images. As input, FusionRF accepts the original multispectral and panchromatic data, eliminating the need for image preprocessing. FusionRF also leverages multimodal appearance embeddings that encode the image characteristics of each modality and view within a uniform representation. By optimizing on both modalities, FusionRF learns to fuse image modalities while performing reconstruction tasks and eliminates the need for a pansharpening preprocessing step. We evaluate our method on multispectral and panchromatic satellite images from the WorldView-3 satellite in various locations, and show that FusionRF provides an average of 17% reduction in depth reconstruction error, and renders sharp training and novel views.

Yutao Tang, Cheng Zhao, Gaurav Mittal et al.

Recent advances in 3D vision-language models (VLMs) highlight a strong potential for 3D scene understanding and reasoning. However, effectively tokenizing 3D scenes into holistic scene tokens, and leveraging these tokens across diverse 3D understanding tasks, remain highly challenging. We present NDTokenizer3D, a generalist 3D VLM that performs a wide range of 3D scene understanding tasks while naturally supporting human interactions, thereby bridging language-level reasoning with 3D spatial understanding. The core of our approach is a novel three-stage scene tokenization pipeline built upon a Multi-Scale Normal Distributions Transform (NDT) representation, paired with a Multi-Scale NDT Decoder (MSDec). Specifically, NDTokenizer3D first constructs a multi-scale NDT representation from raw high-resolution point clouds, preserving both global context and fine-grained geometric details. Next, the MSDec progressively fuses cross-scale NDT features, producing holistic scene tokens consumable by LLM endpoints. Beyond tokenization, MSDec is repurposed as a general interface for human-interactive prompting (points, boxes, masks) and segmentation-mask decoding, unifying diverse 3D scene understanding tasks within a single architecture. With this compact and unified design, NDTokenizer3D offers a fine-grained, general-purpose 3D VLM, achieving remarkable improvements in 3D Referring Segmentation, 3D Visual Question Answering, and 3D Dense Captioning.

Tyler Skow, Alexander Martin, Benjamin Van Durme et al.

Reranking is a critical component of modern retrieval systems, which typically pair an efficient first-stage retriever with a more expressive model to refine results. While large reasoning models have driven rapid progress in text-centric reranking, reasoning-based reranking for video retrieval remains underexplored. To address this gap, we introduce RANKVIDEO, a reasoning-based reranker for video retrieval that explicitly reasons over query-video pairs using video content to assess relevance. RANKVIDEO is trained using a two-stage curriculum consisting of perception-grounded supervised fine-tuning followed by reranking training that combines pointwise, pairwise, and teacher confidence distillation objectives, and is supported by a data synthesis pipeline for constructing reasoning-intensive query-video pairs. Experiments on the large-scale MultiVENT 2.0 benchmark demonstrate that RANKVIDEO consistently improves retrieval performance within a two-stage framework, yielding an average improvement of 31% on nDCG@10 and outperforming text-only and vision-language reranking alternatives, while more efficient.

Anirudh Nanduri, Rama Chellappa

Despite the remarkable performance of deep neural networks for face detection and recognition tasks in the visible spectrum, their performance on more challenging non-visible domains is comparatively still lacking. While significant research has been done in the fields of domain adaptation and domain generalization, in this paper we tackle scenarios in which these methods have limited applicability owing to the lack of training data from target domains. We focus on the problem of single-source (visible) and multi-target (SWIR, long-range/remote, surveillance, and body-worn) face recognition task. We show through experiments that a good template generation algorithm becomes crucial as the complexity of the target domain increases. In this context, we introduce a template generation algorithm called Norm Pooling (and a variant known as Sparse Pooling) and show that it outperforms average pooling across different domains and networks, on the IARPA JANUS Benchmark Multi-domain Face (IJB-MDF) dataset.

Prakhar Kaushik, Ankit Vaidya, Shravan Chaudhari et al.

Adapting large pretrained models to new tasks efficiently and continually is crucial for real-world deployment but remains challenging due to catastrophic forgetting and the high cost of retraining. While parameter-efficient tuning methods like low rank adaptation (LoRA) reduce computational demands, they lack mechanisms for strict continual learning and knowledge integration, without relying on data replay, or multiple adapters. We propose Share, a novel approach to parameter efficient continual finetuning that learns and dynamically updates a single, shared low-rank subspace, enabling seamless adaptation across multiple tasks and modalities. Share constructs a foundational subspace that extracts core knowledge from past tasks and incrementally integrates new information by identifying essential subspace directions. Knowledge from each new task is incorporated into this evolving subspace, facilitating forward knowledge transfer, while minimizing catastrophic interference. This approach achieves up to 100x parameter reduction and 281x memory savings over traditional LoRA methods, maintaining performance comparable to jointly trained models. A single Share model can replace hundreds of task-specific LoRA adapters, supporting scalable, asynchronous continual learning. Experiments across image classification, natural language understanding, 3D pose estimation, and text-to-image generation validate its effectiveness, making Share a practical and scalable solution for lifelong learning in large-scale AI systems.

Aniket Roy, Anshul Shah, Ketul Shah et al.

Visual recognition in a low-data regime is challenging and often prone to overfitting. To mitigate this issue, several data augmentation strategies have been proposed. However, standard transformations, e.g., rotation, cropping, and flipping provide limited semantic variations. To this end, we propose Cap2Aug, an image-to-image diffusion model-based data augmentation strategy using image captions as text prompts. We generate captions from the limited training images and using these captions edit the training images using an image-to-image stable diffusion model to generate semantically meaningful augmentations. This strategy generates augmented versions of images similar to the training images yet provides semantic diversity across the samples. We show that the variations within the class can be captured by the captions and then translated to generate diverse samples using the image-to-image diffusion model guided by the captions. However, naive learning on synthetic images is not adequate due to the domain gap between real and synthetic images. Thus, we employ a maximum mean discrepancy (MMD) loss to align the synthetic images to the real images for minimizing the domain gap. We evaluate our method on few-shot and long-tail classification tasks and obtain performance improvements over state-of-the-art, especially in the low-data regimes.

Aniket Roy, Maitreya Suin, Rama Chellappa

Recent advances in text-to-image diffusion models have substantially improved the quality of image customization, enabling the synthesis of highly realistic images. Despite this progress, achieving fast and efficient personalization remains a key challenge, particularly for real-world applications. Existing approaches primarily accelerate customization for human subjects by injecting identity-specific embeddings into diffusion models, but these strategies do not generalize well to arbitrary object categories, limiting their applicability. To address this limitation, we propose a novel framework that employs a learned network to predict object-specific textual inversion embeddings, which are subsequently integrated into the UNet timesteps of a diffusion model for text-conditional customization. This design enables rapid, zero-shot personalization of a wide range of objects in a single forward pass, offering both flexibility and scalability. Extensive experiments across multiple tasks and settings demonstrate the effectiveness of our approach, highlighting its potential to support fast, versatile, and inclusive image customization. To the best of our knowledge, this work represents the first attempt to achieve such general-purpose, training-free personalization within diffusion models, paving the way for future research in personalized image generation.

Namitha Guruprasad, Abhay Yadav, Cheng Peng et al.

Recovering dense 3D geometry from unposed images remains a foundational challenge in computer vision. Current state-of-the-art models are predominantly trained on perspective datasets, which implicitly constrains them to a standard pinhole camera geometry. As a result, these models suffer from significant geometric degradation when applied to wide-angle imagery captured via non-rectilinear optics, such as fisheye or panoramic sensors. To address this, we present CAM3R, a Camera-Agnostic, feed-forward Model for 3D Reconstruction capable of processing images from wide-angle camera models without prior calibration. Our framework consists of a two-view network which is bifurcated into a Ray Module (RM) to estimate per-pixel ray directions and a Cross-view Module (CVM) to infer radial distance with confidence maps, pointmaps, and relative poses. To unify these pairwise predictions into a consistent 3D scene, we introduce a Ray-Aware Global Alignment framework for pose refinement and scale optimization while strictly preserving the predicted local geometry. Extensive experiments on various camera model datasets, including panorama, fisheye and pinhole imagery, demonstrate that CAM3R establishes a new state-of-the-art in pose estimation and reconstruction.

Chandrakanth Gudavalli, Tajuddin Manhar Mohammed, Abhay Yadav et al.

Aligning ground-level imagery with geo-registered satellite maps is crucial for mapping, navigation, and situational awareness, yet remains challenging under large viewpoint gaps or when GPS is unreliable. We introduce Wrivinder, a zero-shot, geometry-driven framework that aggregates multiple ground photographs to reconstruct a consistent 3D scene and align it with overhead satellite imagery. Wrivinder combines SfM reconstruction, 3D Gaussian Splatting, semantic grounding, and monocular depth--based metric cues to produce a stable zenith-view rendering that can be directly matched to satellite context for metrically accurate camera geo-localization. To support systematic evaluation of this task, which lacks suitable benchmarks, we also release MC-Sat, a curated dataset linking multi-view ground imagery with geo-registered satellite tiles across diverse outdoor environments. Together, Wrivinder and MC-Sat provide a first comprehensive baseline and testbed for studying geometry-centered cross-view alignment without paired supervision. In zero-shot experiments, Wrivinder achieves sub-30\,m geolocation accuracy across both dense and large-area scenes, highlighting the promise of geometry-based aggregation for robust ground-to-satellite localization.

Xing Han, Shravan Chaudhari, Tanvi Ranade et al.

Real-world model deployment across multiple domains requires multimodal models to operate under two complementary regimes: (1) multi-task pretraining, tasks are co-available at design time where related tasks could borrow representational strength from one another, (2) continual adaptation, in which new tasks emerge after deployment with previously unseen modality combinations. However, neither regime alone suffices: the pretraining task set is never exhaustive, while bypassing joint training forfeits the transfer gains and efficiency among co-trainable tasks. Sparse Mixture-of-Experts (MoE) is a natural fit for this dual requirement: sparse activation enables modular capacity expansion as new tasks arrive, while routing decouples modality-level computation from task-level composition. In this work, we propose a scalable MoE framework for multitask pretraining and continual learning across flexible modality combinations. The framework is designed to support training on multimodal tasks with diverse modality configurations by leveraging modality-specific routers that process tokens from each modality across tasks. Furthermore, it enables continual learning over sequential multimodal tasks within a fixed-capacity MoE by compressing accumulated expert knowledge into low-rank memory subspaces, while expanding only the lightweight routers. We validate the effectiveness of our method on multiple healthcare multimodal benchmarks. It demonstrates competitive multitask pretraining performance while alleviating catastrophic forgetting and improving parameter efficiency.

Ayush Gupta, Rama Chellappa · amazon-science

Gait recognition is an important biometric technique over large distances. State-of-the-art gait recognition systems perform very well in controlled environments at close range. Recently, there has been an increased interest in gait recognition in the wild prompted by the collection of outdoor, more challenging datasets containing variations in terms of illumination, pitch angles, and distances. An important problem in these environments is that of occlusion, where the subject is partially blocked from camera view. While important, this problem has received little attention. Thus, we propose MimicGait, a model-agnostic approach for gait recognition in the presence of occlusions. We train the network using a multi-instance correlational distillation loss to capture both inter-sequence and intra-sequence correlations in the occluded gait patterns of a subject, utilizing an auxiliary Visibility Estimation Network to guide the training of the proposed mimic network. We demonstrate the effectiveness of our approach on challenging real-world datasets like GREW, Gait3D and BRIAR. We release the code in https://github.com/Ayush-00/mimicgait.

Hossein Souri, Arpit Bansal, Hamid Kazemi et al.

Modern neural networks are often trained on massive datasets that are web scraped with minimal human inspection. As a result of this insecure curation pipeline, an adversary can poison or backdoor the resulting model by uploading malicious data to the internet and waiting for a victim to scrape and train on it. Existing approaches for creating poisons and backdoors start with randomly sampled clean data, called base samples, and then modify those samples to craft poisons. However, some base samples may be significantly more amenable to poisoning than others. As a result, we may be able to craft more potent poisons by carefully choosing the base samples. In this work, we use guided diffusion to synthesize base samples from scratch that lead to significantly more potent poisons and backdoors than previous state-of-the-art attacks. Our Guided Diffusion Poisoning (GDP) base samples can be combined with any downstream poisoning or backdoor attack to boost its effectiveness. Our implementation code is publicly available at: https://github.com/hsouri/GDP .

Siyuan Huang, Yutong Gao, Juyang Bai et al.

Creating high-quality figures and visualizations for scientific papers is a time-consuming task that requires both deep domain knowledge and professional design skills. Despite over 2.5 million scientific papers published annually, the figure generation process remains largely manual. We introduce $\textbf{SciFig}$, an end-to-end AI agent system that generates publication-ready pipeline figures directly from research paper texts. SciFig uses a hierarchical layout generation strategy, which parses research descriptions to identify component relationships, groups related elements into functional modules, and generates inter-module connections to establish visual organization. Furthermore, an iterative chain-of-thought (CoT) feedback mechanism progressively improves layouts through multiple rounds of visual analysis and reasoning. We introduce a rubric-based evaluation framework that analyzes 2,219 real scientific figures to extract evaluation rubrics and automatically generates comprehensive evaluation criteria. SciFig demonstrates remarkable performance: achieving 70.1$\%$ overall quality on dataset-level evaluation and 66.2$\%$ on paper-specific evaluation, and consistently high scores across metrics such as visual clarity, structural organization, and scientific accuracy. SciFig figure generation pipeline and our evaluation benchmark will be open-sourced.