Computer Vision · Cross-listed from cs.LG
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
Sunday 23 November
We propose TimePre, a simple framework that unifies the efficiency of Multilayer Perceptron (MLP)-based models with the distributional flexibility of Multiple Choice Learning (MCL) for Probabilistic Time-Series Forecasting (PTSF). Stabilized Instance Normalization (SIN), the core of TimePre, is a normalization layer that explicitly addresses the trade-off among accuracy, efficiency, and stability. SIN stabilizes the hybrid architecture by correcting channel-wise statistical shifts, thereby resolving the catastrophic hypothesis collapse. Extensive experiments on six benchmark datasets demonstrate that TimePre achieves state-of-the-art (SOTA) accuracy on key probabilistic metrics. Critically, TimePre achieves inference speeds that are orders of magnitude faster than sampling-based models, and is more stable than prior MCL approaches.
Computer Vision
Zero-Shot Video Deraining with Video Diffusion Models
Existing video deraining methods are often trained on paired datasets, either synthetic, which limits their ability to generalize to real-world rain, or captured by static cameras, which restricts their effectiveness in dynamic scenes with background and camera motion. Furthermore, recent works in fine-tuning diffusion models have shown promising results, but the fine-tuning tends to weaken the generative prior, limiting generalization to unseen cases. In this paper, we introduce the first zero-shot video deraining method for complex dynamic scenes that does not require synthetic data nor model fine-tuning, by leveraging a pretrained text-to-video diffusion model that demonstrates strong generalization capabilities. By inverting an input video into the latent space of diffusion models, its reconstruction process can be intervened and pushed away from the model's concept of rain using negative prompting. At the core of our approach is an attention switching mechanism that we found is crucial for maintaining dynamic backgrounds as well as structural consistency between the input and the derained video, mitigating artifacts introduced by naive negative prompting. Our approach is validated through extensive experiments on real-world rain datasets, demonstrating substantial improvements over prior methods and showcasing robust generalization without the need for supervised training.
Read the full abstract
Computer Vision
HiFi-MambaV2: Hierarchical Shared-Routed MoE for High-Fidelity MRI Reconstruction
Reconstructing high-fidelity MR images from undersampled k-space data requires recovering high-frequency details while maintaining anatomical coherence. We present HiFi-MambaV2, a hierarchical shared-routed Mixture-of-Experts (MoE) Mamba architecture that couples frequency decomposition with content-adaptive computation. The model comprises two core components: (i) a separable frequency-consistent Laplacian pyramid (SF-Lap) that delivers alias-resistant, stable low- and high-frequency streams; and (ii) a hierarchical shared-routed MoE that performs per-pixel top-1 sparse dispatch to shared experts and local routers, enabling effective specialization with stable cross-depth behavior. A lightweight global context path is fused into an unrolled, data-consistency-regularized backbone to reinforce long-range reasoning and preserve anatomical coherence. Evaluated on fastMRI, CC359, ACDC, M4Raw, and Prostate158, HiFi-MambaV2 consistently outperforms CNN-, Transformer-, and prior Mamba-based baselines in PSNR, SSIM, and NMSE across single- and multi-coil settings and multiple acceleration factors, consistently surpassing consistent improvements in high-frequency detail and overall structural fidelity. These results demonstrate that HiFi-MambaV2 enables reliable and robust MRI reconstruction.
Read the full abstract
Computer Vision
DE-KAN: A Kolmogorov Arnold Network with Dual Encoder for accurate 2D Teeth Segmentation
Accurate segmentation of individual teeth from panoramic radiographs remains a challenging task due to anatomical variations, irregular tooth shapes, and overlapping structures. These complexities often limit the performance of conventional deep learning models. To address this, we propose DE-KAN, a novel Dual Encoder Kolmogorov Arnold Network, which enhances feature representation and segmentation precision. The framework employs a ResNet-18 encoder for augmented inputs and a customized CNN encoder for original inputs, enabling the complementary extraction of global and local spatial features. These features are fused through KAN-based bottleneck layers, incorporating nonlinear learnable activation functions derived from the Kolmogorov Arnold representation theorem to improve learning capacity and interpretability. Extensive experiments on two benchmark dental X-ray datasets demonstrate that DE-KAN outperforms state-of-the-art segmentation models, achieving mIoU of 94.5%, Dice coefficient of 97.1%, accuracy of 98.91%, and recall of 97.36%, representing up to +4.7% improvement in Dice compared to existing methods.
Read the full abstract
Computer Vision · Cross-listed from cs.RO
Splatblox: Traversability-Aware Gaussian Splatting for Outdoor Robot Navigation
We present Splatblox, a real-time system for autonomous navigation in outdoor environments with dense vegetation, irregular obstacles, and complex terrain. Our method fuses segmented RGB images and LiDAR point clouds using Gaussian Splatting to construct a traversability-aware Euclidean Signed Distance Field (ESDF) that jointly encodes geometry and semantics. Updated online, this field enables semantic reasoning to distinguish traversable vegetation (e.g., tall grass) from rigid obstacles (e.g., trees), while LiDAR ensures 360-degree geometric coverage for extended planning horizons. We validate Splatblox on a quadruped robot and demonstrate transfer to a wheeled platform. In field trials across vegetation-rich scenarios, it outperforms state-of-the-art methods with over 50% higher success rate, 40% fewer freezing incidents, 5% shorter paths, and up to 13% faster time to goal, while supporting long-range missions up to 100 meters. Experiment videos and more details can be found on our project page: https://splatblox.github.io
Read the full abstract
Computer Vision
Breaking Forgetting: Training-Free Few-Shot Class-Incremental Learning via Conditional Diffusion
Efforts to overcome catastrophic forgetting in Few-Shot Class-Incremental Learning (FSCIL) have primarily focused on developing more effective gradient-based optimization strategies. In contrast, little attention has been paid to the training cost explosion that inevitably arises as the number of novel classes increases, a consequence of relying on gradient learning even under extreme data scarcity. More critically, since FSCIL typically provides only a few samples for each new class, gradient-based updates not only induce severe catastrophic forgetting on base classes but also hinder adaptation to novel ones. This paper seeks to break this long-standing limitation by asking: Can we design a training-free FSCIL paradigm that entirely removes gradient optimization? We provide an affirmative answer by uncovering an intriguing connection between gradient-based optimization and the Conditional Diffusion process. Building on this observation, we propose a Conditional Diffusion-driven FSCIL (CD-FSCIL) framework that substitutes the conventional gradient update process with a diffusion-based generative transition, enabling training-free incremental adaptation while effectively mitigating forgetting. Furthermore, to enhance representation under few-shot constraints, we introduce a multimodal learning strategy that integrates visual features with natural language descriptions automatically generated by Large Language Models (LLMs). This synergy substantially alleviates the sample scarcity issue and improves generalization across novel classes. Extensive experiments on mainstream FSCIL benchmarks demonstrate that our method not only achieves state-of-the-art performance but also drastically reduces computational and memory overhead, marking a paradigm shift toward training-free continual adaptation.
Read the full abstract
Computer Vision
Unified Deep Learning Platform for Dust and Fault Diagnosis in Solar Panels Using Thermal and Visual Imaging
Solar energy is one of the most abundant and tapped sources of renewable energies with enormous future potential. Solar panel output can vary widely with factors like intensity, temperature, dirt, debris and so on affecting it. We have implemented a model on detecting dust and fault on solar panels. These two applications are centralized as a single-platform and can be utilized for routine-maintenance and any other checks. These are checked against various parameters such as power output, sinusoidal wave (I-V component of solar cell), voltage across each solar cell and others. Firstly, we filter and preprocess the obtained images using gamma removal and Gaussian filtering methods alongside some predefined processes like normalization. The first application is to detect whether a solar cell is dusty or not based on various pre-determined metrics like shadowing, leaf, droppings, air pollution and from other human activities to extent of fine-granular solar modules. The other one is detecting faults and other such occurrences on solar panels like faults, cracks, cell malfunction using thermal imaging application. This centralized platform can be vital since solar panels have different efficiency across different geography (air and heat affect) and can also be utilized for small-scale house requirements to large-scale solar farm sustentation effectively. It incorporates CNN, ResNet models that with self-attention mechanisms-KerNet model which are used for classification and results in a fine-tuned system that detects dust or any fault occurring. Thus, this multi-application model proves to be efficient and optimized in detecting dust and faults on solar panels. We have performed various comparisons and findings that demonstrates that our model has better efficiency and accuracy results overall than existing models.
Read the full abstract
Computer Vision
LRDUN: A Low-Rank Deep Unfolding Network for Efficient Spectral Compressive Imaging
Deep unfolding networks (DUNs) have achieved remarkable success and become the mainstream paradigm for spectral compressive imaging (SCI) reconstruction. Existing DUNs are derived from full-HSI imaging models, where each stage operates directly on the high-dimensional HSI, refining the entire data cube based on the single 2D coded measurement. However, this paradigm leads to computational redundancy and suffers from the ill-posed nature of mapping 2D residuals back to 3D space of HSI. In this paper, we propose two novel imaging models corresponding to the spectral basis and subspace image by explicitly integrating low-rank (LR) decomposition with the sensing model. Compared to recovering the full HSI, estimating these compact low-dimensional components significantly mitigates the ill-posedness. Building upon these novel models, we develop the Low-Rank Deep Unfolding Network (LRDUN), which jointly solves the two subproblems within an unfolded proximal gradient descent (PGD) framework. Furthermore, we introduce a Generalized Feature Unfolding Mechanism (GFUM) that decouples the physical rank in the data-fidelity term from the feature dimensionality in the prior module, enhancing the representational capacity and flexibility of the network. Extensive experiments on simulated and real datasets demonstrate that the proposed LRDUN achieves state-of-the-art (SOTA) reconstruction quality with significantly reduced computational cost.
Read the full abstract
Computer Vision
Multimodal Continual Learning with MLLMs from Multi-scenario Perspectives
Multimodal large language models (MLLMs) deployed on devices must adapt to continuously changing visual scenarios such as variations in background and perspective, to effectively perform complex visual tasks. To investigate catastrophic forgetting under real-world scenario shifts, we construct a multimodal visual understanding dataset (MSVQA), covering four distinct scenarios and perspectives: high-altitude, underwater, low-altitude, and indoor environments. Furthermore, we propose UNIFIER (mUltimodal coNtInual learning with MLLMs From multi-scenarIo pERspectives), a continual learning (CL) framework designed to address visual discrepancies while learning different scenarios. Compared to existing CL methods, UNIFIER enables knowledge accumulation within the same scenario and mutual enhancement across different scenarios via Vision Representation Expansion (VRE) and Vision Consistency Constraint (VCC). Experimental results show that UNIFIER improves the last-step VQA scores by 2.70%~10.62% and the last-step F1 scores by 3.40%~7.69% compared to the state-of-the-art method, QUAD, in 20-step cross-scenario continual learning tasks. MSVQA dataset is available at https://huggingface.co/datasets/Kaij00/MSVQA.
Read the full abstract
Computer Vision
Extreme Model Compression for Edge Vision-Language Models: Sparse Temporal Token Fusion and Adaptive Neural Compression
The demand for edge AI in vision-language tasks requires models that achieve real-time performance on resource-constrained devices with limited power and memory. This paper proposes two adaptive compression techniques – Sparse Temporal Token Fusion (STTF) and Adaptive Neural Compression (ANC) – that integrate algorithmic innovations with hardware-aware optimizations. Unlike previous approaches relying on static pruning or uniform scaling, STTF dynamically reuses visual tokens through event-driven change detection, while ANC conditionally activates encoder branches via a learned router, enabling fine-grained adaptation to scene complexity. Our 3B-parameter TinyGPT-STTF achieves CIDEr 131.2, BLEU-4 0.38, METEOR 0.31, and ROUGE-L 0.56 on the COCO 2017 test set, surpassing LLaVA-1.5 7B by 17.6 CIDEr points while using 2.3x fewer parameters and 62x fewer on-device FLOPs. TinyGPT-ANC reaches CIDEr 128.5. On event-based vision tasks, STTF reduces average token count by 84% (from 196 to 31 tokens) while preserving 95.6% accuracy on the DVS128 Gesture dataset, and ANC cuts FLOPs by up to 90% in low-motion scenes. Compared to strong baselines, our models improve accuracy by up to 4.4% and reduce latency by up to 13x. These results enable efficient deployment of capable vision-language models on real-world edge devices.
Read the full abstract
Computer Vision · Cross-listed from eess.IV
SAGE: Shape-Adapting Gated Experts for Adaptive Histopathology Image Segmentation
The significant variability in cell size and shape continues to pose a major obstacle in computer-assisted cancer detection on gigapixel Whole Slide Images (WSIs), due to cellular heterogeneity. Current CNN-Transformer hybrids use static computation graphs with fixed routing. This leads to extra computation and makes it harder to adapt to changes in input. We propose Shape-Adapting Gated Experts (SAGE), an input-adaptive framework that enables dynamic expert routing in heterogeneous visual networks. SAGE reconfigures static backbones into dynamically routed expert architectures via a dual-path design with hierarchical gating and a Shape-Adapting Hub (SA-Hub) that harmonizes feature representations across convolutional and transformer modules. Embodied as SAGE with ConvNeXt and Vision Transformer UNet (SAGE-ConvNeXt+ViT-UNet), our model achieves a Dice score of 95.23% on EBHI, DSC scores of 92.78% and 91.42% on GlaS Test A and Test B, respectively, and 91.26% DSC at the WSI level on DigestPath, while exhibiting robust generalization under distribution shifts by adaptively balancing local refinement and global context. SAGE establishes a scalable foundation for dynamic expert routing in visual networks, thereby facilitating flexible visual reasoning. Project page: https://oxyzgiahuy.github.io/sage/
Read the full abstract
Computer Vision
Uncertainty Quantification in HSI Reconstruction using Physics-Aware Diffusion Priors and Optics-Encoded Measurements
Hyperspectral image reconstruction from a compressed measurement is a highly ill-posed inverse problem. Current data-driven methods suffer from hallucination due to the lack of spectral diversity in existing hyperspectral image datasets, particularly when they are evaluated for the metamerism phenomenon. In this work, we formulate hyperspectral image (HSI) reconstruction as a Bayesian inference problem and propose a framework, HSDiff, that utilizes an unconditionally trained, pixel-level diffusion prior and posterior diffusion sampling to generate diverse HSI samples consistent with the measurements of various hyperspectral image formation models. We propose an enhanced metameric augmentation technique using region-based metameric black and partition-of-union spectral upsampling to expand training with physically valid metameric spectra, strengthening the prior diversity and improving uncertainty calibration. We utilize HSDiff to investigate how the studied forward models shape the posterior distribution and demonstrate that guiding with effective spectral encoding provides calibrated informative uncertainty compared to non-encoded models. Through the lens of the Bayesian framework, HSDiff offers a complete, high-performance method for uncertainty-aware HSI reconstruction. Our results also reiterate the significance of effective spectral encoding in snapshot hyperspectral imaging.
Read the full abstract
Computer Vision
Jacobian-Aware Posterior Sampling for Inverse Problems
Diffusion models provide powerful generative priors for solving inverse problems by sampling from a posterior distribution conditioned on corrupted measurements. Existing methods primarily follow two paradigms: direct methods, which approximate the likelihood term, and proximal methods, which incorporate intermediate solutions satisfying measurement constraints into the sampling process. Under standard Gaussian approximations and locally-linear measurements, we demonstrate that these approaches differ fundamentally in their treatment of the diffusion denoiser's Jacobian within the likelihood term. While this Jacobian encodes critical prior knowledge of the data distribution, training-induced non-idealities can degrade performance in zero-shot settings. In this work, we bridge direct and proximal approaches by proposing a principled Jacobian-Aware Posterior Sampler (JAPS). JAPS leverages the Jacobian's prior knowledge while mitigating its detrimental effects through a corresponding proximal solution, requiring no additional computational cost. Additionally, we integrate our guidance into DDIM sampling, with a corrected conditional factor that has been missing in previous works. Our method enhances reconstruction quality across diverse linear and nonlinear noisy imaging tasks, outperforming existing diffusion-based baselines in perceptual quality while maintaining or improving distortion metrics.
Read the full abstract
Computer Vision
Gaze Beyond the Frame: Forecasting Egocentric 3D Visual Span
People continuously perceive and interact with their surroundings based on underlying intentions that drive their exploration and behaviors. While research in egocentric user and scene understanding has focused primarily on motion and contact-based interaction, forecasting human visual perception itself remains less explored despite its fundamental role in guiding human actions and its implications for AR/VR and assistive technologies. We address the challenge of egocentric 3D visual span forecasting, predicting where a person's visual perception will focus next within their three-dimensional environment. To this end, we propose EgoSpanLift, a novel method that transforms egocentric visual span forecasting from 2D image planes to 3D scenes. EgoSpanLift converts SLAM-derived keypoints into gaze-compatible geometry and extracts volumetric visual span regions. We further combine EgoSpanLift with 3D U-Net and unidirectional transformers, enabling spatio-temporal fusion to efficiently predict future visual span in the 3D grid. In addition, we curate a comprehensive benchmark from raw egocentric multisensory data, creating a testbed with 364.6K samples for 3D visual span forecasting. Our approach outperforms competitive baselines for egocentric 2D gaze anticipation and 3D localization while achieving comparable results even when projected back onto 2D image planes without additional 2D-specific training.
Read the full abstract
Computer Vision · Cross-listed from cs.LG
SloMo-Fast: Slow-Momentum and Fast-Adaptive Teachers for Source-Free Continual Test-Time Adaptation
Continual Test-Time Adaptation (CTTA) is crucial for deploying models in real-world applications with unseen, evolving target domains. Existing CTTA methods, however, often rely on source data or prototypes, limiting their applicability in privacy-sensitive and resource-constrained settings. Although several methods attempt to mitigate catastrophic forgetting, they often fail to preserve long-term domain-specific knowledge across many domain shifts. Moreover, their relatively slow adaptation rates during domain transitions can cause error accumulation, allowing mistakes to propagate before effective adaptation occurs. To address these challenges, we propose SloMo-Fast, a source-free, dual-teacher CTTA framework designed for enhanced quick adaptability and generalization. It includes two complementary teachers: the Slow-Teacher, which exhibits slow forgetting and retains long-term knowledge of previously encountered domains to ensure robust generalization, and the Fast-Teacher rapidly adapts to new domains while accumulating and integrating knowledge across them. This framework preserves knowledge of past domains and adapts efficiently to new ones. Our extensive experiments show that SloMo-Fast consistently outperforms state-of-the-art methods across Cyclic Test-Time Adaptation (Cyclic-TTA), a CTTA benchmark that simulates recurring domain shifts, along with ten other CTTA settings, highlighting its ability to both adapt and generalize across evolving, revisited domains.
Read the full abstract
Computer Vision
Decoupling Perception from Reasoning for Hallucination-Resistant Video Understanding
Video Large Language Models improve reasoning over complex videos by generating intermediate reasoning text. However, reliable reasoning depends on accurate video perception. In existing approaches, perception evidence is intertwined with reasoning text, making it difficult to directly supervise the perception process. We argue that reliable supervision requires explicitly separating perception evidence from reasoning so that perception can be verified independently. To supervise perception directly, we propose Decoupled Perception and Logic (DPL), which represents perception as fixed-format evidence units containing timestamps and visual descriptions. This structured representation enables direct extraction of perception content and simplifies alignment between video segments and reward evaluation. Building on DPL, we introduce a perception reward that encourages both hallucination resistance and perception-based reasoning. An Factual-Aware Evaluator (FAE) provides anti-hallucination scores and achieves hallucination evaluation performance comparable to GPT-4o. In addition, we validate reasoning consistency by feeding perception results and questions into a reference model. Experiments show that, by providing reliable process rewards, Video-DPL consistently improves post-training performance at both 3B and 7B scales, while delivering higher data efficiency.
Read the full abstract
Computer Vision · Cross-listed from cs.LG
Radiation-Preserving Selective Imaging for Pediatric Hip Dysplasia: A Cross-Modal Ultrasound-Xray Policy with Limited Labels
We study an ultrasound-first, radiation-preserving policy for developmental dysplasia of the hip (DDH) that requests a radiograph only when needed. We (i) pretrain modality-specific encoders (ResNet-18) with SimSiam on a large unlabelled registry (37186 ultrasound; 19546 radiographs), (ii) freeze the backbones and fit small, measurement-faithful heads on DDH-relevant landmarks and measurements, (iii) calibrate a one-sided conformal deferral rule on ultrasound predictions that provides finite sample marginal coverage guarantees under exchangeability, using a held-out calibration set. Ultrasound heads predict Graf alpha, beta, and femoral head coverage; X-ray heads predict acetabular index (AI), center-edge (CE) angle and IHDI grade. On our held out labeled evaluation set, ultrasound measurement error is modest (e.g., alpha MAE ~= 9.7 degrees, coverage MAE ~= 14.0%), while radiographic probes achieve AI and CE MAEs of ~= 7.6 degrees and ~= 8.9 degrees, respectively. The calibrated US-only policy is explored across rule families (alpha-only; alpha OR coverage; alpha AND coverage), conformal miscoverage levels, and per-utility trade-offs using decision-curve analysis. Conservative settings yield high coverage with near-zero US-only rates; permissive settings (e.g., alpha OR coverage at larger deltas) achieve non-zero US-only throughput with expected coverage tradeoffs. The result is a simple, reproducible pipeline that turns limited labels into interpretable measurements and tunable selective imaging curves suitable for clinical handoff and future external validation.
Read the full abstract
Computer Vision
AttnRegDeepLab: A Two-Stage Decoupled Framework for Interpretable Embryo Fragmentation Grading
Assessing embryo fragmentation is crucial for predicting IVF success, yet manual grading is prone to subjectivity, and existing AI models struggle with clinical interpretability and segmentation errors. We propose AttnRegDeepLab, a Multi-Task Learning (MTL) framework designed to solve these challenges. The model enhances a DeepLabV3+ decoder with Attention Gates to filter out cytoplasmic noise and retain sharp contour details. It also introduces a Multi-Scale Regression Head with Feature Injection, guiding the segmentation process with global grading priors to eliminate systematic area estimation errors. Based on a two-stage decoupled training strategy and a range-based loss for weakly labeled data, our method resolves MTL gradient conflicts. AttnRegDeepLab yields high grading precision and excellent segmentation quality (Dice coefficient = 0.729), avoiding the trade-off between contour integrity and grading accuracy seen under standard joint optimization. This provides a reliable, clinically interpretable tool balancing visual and quantitative accuracy.
Read the full abstract
Computer Vision
NAF: Zero-Shot Feature Upsampling via Neighborhood Attention Filtering
Vision Foundation Models (VFMs) extract spatially downsampled representations, posing challenges for pixel-level tasks. Existing upsampling approaches face a fundamental trade-off: classical filters are fast and broadly applicable but rely on fixed forms, while modern upsamplers achieve superior accuracy through learnable, VFM-specific forms at the cost of retraining for each VFM. We introduce Neighborhood Attention Filtering (NAF), which bridges this gap by learning adaptive spatial-and-content weights through Cross-Scale Neighborhood Attention and Rotary Position Embeddings (RoPE), guided solely by the high-resolution input image. NAF operates zero-shot: it upsamples features from any VFM without retraining, making it the first VFM-agnostic architecture to outperform VFM-specific upsamplers and achieve state-of-the-art performance across multiple downstream tasks. It maintains high efficiency, scaling to 2K feature maps and reconstructing intermediate-resolution maps at 18 FPS. Beyond feature upsampling, NAF demonstrates strong performance on image restoration, highlighting its versatility. Code and checkpoints are available at https://github.com/valeoai/NAF.
Read the full abstract
Computer Vision
EventBench: Towards Comprehensive Benchmarking of Event-based MLLMs
Multimodal large language models (MLLMs) have made significant advancements in event-based vision, yet the comprehensive evaluation of their capabilities within a unified benchmark remains largely unexplored. In this work, we introduce EventBench, a benchmark that offers eight diverse task metrics together with a large-scale event stream dataset. EventBench differs from existing event-based benchmarks in four key aspects: (1) openness in accessibility, releasing all raw event streams and task instructions across eight evaluation metrics; (2) diversity in task coverage, spanning understanding, recognition, and spatial reasoning tasks for comprehensive capability assessment; (3) integration in spatial dimensions, pioneering the design of 3D spatial reasoning tasks for event-based MLLMs; and (4) scale in data volume, with an accompanying training set of over one million event-text pairs supporting large-scale training and evaluation. Using EventBench, we evaluate state-of-the-art closed-source models such as GPT-5 and Gemini-2.5 Pro, leading open-source models including Qwen2.5-VL and InternVL3, and event-based MLLMs such as EventGPT that directly process raw event streams. Extensive evaluation reveals that while current event-based MLLMs demonstrate strong performance in event stream understanding, they continue to struggle with fine-grained recognition and spatial reasoning.
Read the full abstract
Computer Vision
SineProject: Machine Unlearning for Stable Vision Language Alignment
Multimodal Large Language Models (MLLMs) increasingly need to forget specific knowledge such as unsafe or private information without requiring full retraining. However, existing unlearning methods often disrupt vision language alignment, causing models to reject both harmful and benign queries. We trace this failure to the projector network during unlearning, its Jacobian becomes severely illconditioned, leading to unstable optimization and drift in cross modal embeddings. We introduce SineProject, a simple method that augments the frozen projector with sinusoidally modulated trainable parameters, improving the Jacobian's spectral conditioning and stabilizing alignment throughout unlearning. Across standard safety and privacy unlearning benchmarks using LLaVA v1.5 7B and 13B, SineProject reduces benign query refusals while achieving complete forgetting of targeted information, yielding state of the art forget retain trade offs with negligible computational overhead.
Read the full abstract
Computer Vision
ReCoGS: Real-time ReColoring for Gaussian Splatting scenes
Gaussian Splatting has emerged as a leading method for novel view synthesis, offering superior training efficiency and real-time inference compared to NeRF approaches, while still delivering high-quality reconstructions. Beyond view synthesis, this 3D representation has also been explored for editing tasks. Many existing methods leverage 2D diffusion models to generate multi-view datasets for training, but they often suffer from limitations such as view inconsistencies, lack of fine-grained control, and high computational demand. In this work, we focus specifically on the editing task of recoloring. We introduce a user-friendly pipeline that enables precise selection and recoloring of regions within a pre-trained Gaussian Splatting scene. To demonstrate the real-time performance of our method, we also present an interactive tool that allows users to experiment with the pipeline in practice. Code is available at https://github.com/loryruta/recogs.
Read the full abstract
Computer Vision
Perceptual-Evidence Anchored Reinforced Learning for Multimodal Reasoning
Reinforcement Learning with Verifiable Rewards (RLVR) has significantly advanced the reasoning capabilities of Large Language Models (LLMs) and is now being applied to Vision-Language Models (VLMs). However, vanilla RLVR for VLMs verifies only the final textual output, critically neglecting the foundational step of visual perception. This oversight leads to visual hallucinations and reward hacking, as reasoning built upon flawed perception is inherently unreliable. To address this, we propose PEARL (Perceptual-Evidence Anchored Reinforced Learning), a dual-branch, perception-reasoning synergistic that strengthens multimodal reasoning by explicitly anchoring it to verified visual evidence. For each reasoning-oriented QA instance, PEARL first derive a perception checklist – a set of perception-oriented sub-questions with verifiable answers that probe the model's understanding of key visual evidence. During training, auxiliary rollouts on this checklist yield a perceptual reward that both directly reinforces the model's perception ability and acts as a fidelity gate for reasoning. If the model passes the perception check, its policy update is biased towards evidence-anchored reasoning. Otherwise, the process is halted to prevent reasoning from flawed premises. PEARL can be seamlessly integrated with popular RL methods like GRPO and DAPO. Comprehensive experiments show PEARL achieves substantial gains on multimodal reasoning benchmarks, e.g., a +9.7% improvement over the baseline and +6.6% over GRPO on MathVerse.
Read the full abstract
Computer Vision
When Generative Replay Meets Evolving Deepfakes: Dual Confusion-Aware Regularization for Incremental Face Forgery Detection
The rapid advancement of face generation techniques has introduced an increasing variety of forgery methods, making incremental deepfake detection essential for maintaining robust detectors. While generative replay provides a promising solution to alleviate catastrophic forgetting without storing historical data, its effectiveness is hindered by domain confusion between generated samples and real data. We observe that replay samples fall into two categories: when the replay generator closely resembles the newly introduced forgery model, generated real samples overlap with the fake domain and become domain-risky; when the generator differs significantly, generated samples maintain clearer domain separation and can be treated as domain-safe. To address this challenge, we propose a Dual Confusion-Aware REgularization strategy, termed Dual-CARE. A Domain-aware Confusion Score (DC Score) is introduced to quantify domain confusion and dual-modulate the optimization of both replay generators and the incremental detector. Guided by DC Score, replay generators are updated to better approximate previous-task distributions, while the detector adopts different supervision strategies: domain-safe samples are directly supervised, whereas domain-risky samples are regulated using a Relative Separation Loss to balance supervision and confusion. Extensive experiments demonstrate that Dual-CARE effectively exploits generative replay and improves incremental deepfake detection under evolving forgery scenarios.
Read the full abstract
Computer Vision
DocPTBench: Benchmarking End-to-End Photographed Document Parsing and Translation
The advent of Multimodal Large Language Models (MLLMs) has unlocked the potential for end-to-end document parsing and translation. However, prevailing benchmarks such as OmniDocBench and DITrans are dominated by pristine scanned or digital-born documents, and thus fail to adequately represent the intricate challenges of real-world capture conditions, such as geometric distortions and photometric variations. To fill this gap, we introduce DocPTBench, a comprehensive benchmark specifically designed for Photographed Document Parsing and Translation. DocPTBench comprises over 1,300 high-resolution photographed documents from multiple domains, includes eight translation scenarios, and provides meticulously human-verified annotations for both parsing and translation. Our experiments demonstrate that transitioning from digital-born to photographed documents results in a substantial performance decline: popular MLLMs exhibit an average accuracy drop of 18% in end-to-end parsing and 12% in translation, while specialized document parsing models show significant average decrease of 25%. This substantial performance gap underscores the unique challenges posed by documents captured in real-world conditions and reveals the limited robustness of existing models. Dataset and code are available at https://github.com/Topdu/DocPTBench.
Read the full abstract
Computer Vision
LungX: A Hybrid EfficientNet-Vision Transformer Architecture with Multi-Scale Attention for Accurate Pneumonia Detection
Pneumonia remains a leading global cause of mortality where timely diagnosis is critical. We introduce LungX, a novel hybrid architecture combining EfficientNet's multi-scale features, CBAM attention mechanisms, and Vision Transformer's global context modeling for enhanced pneumonia detection. Evaluated on 20,000 curated chest X-rays from RSNA and CheXpert, LungX achieves state-of-the-art performance (86.5 percent accuracy, 0.943 AUC), representing a 6.7 percent AUC improvement over EfficientNet-B0 baselines. Visual analysis demonstrates superior lesion localization through interpretable attention maps. Future directions include multi-center validation and architectural optimizations targeting 88 percent accuracy for clinical deployment as an AI diagnostic aid.
Read the full abstract
Computer Vision
CrossJEPA: Cross-Modal Joint-Embedding Predictive Architecture for Efficient 3D Representation Learning from 2D Images
Image-to-point cross-modal learning has emerged to address the scarcity of large-scale 3D datasets in 3D representation learning. However, current methods that leverage 2D data often result in large, slow-to-train models, making them computationally expensive and difficult to deploy in resource-constrained environments. The architecture design of such models is therefore critical, determining their performance, memory footprint, and compute efficiency. The Joint-embedding Predictive Architecture (JEPA) has gained wide popularity in self-supervised learning for its simplicity and efficiency, but has been under-explored in cross-modal settings, partly due to the misconception that masking is intrinsic to JEPA. In this light, we propose CrossJEPA, a simple Cross-modal Joint Embedding Predictive Architecture that harnesses the knowledge of an image foundation model and trains a predictor to infer embeddings of specific rendered 2D views from corresponding 3D point clouds, thereby introducing a JEPA-style pretraining strategy beyond masking. By conditioning the predictor on cross-domain projection information, CrossJEPA purifies the supervision signal from semantics exclusive to the target domain. We further exploit the frozen teacher design with a one-time target embedding caching mechanism, yielding amortized efficiency. CrossJEPA achieves a new state-of-the-art in linear probing on the synthetic ModelNet40 (94.2%) and the real-world ScanObjectNN (88.3%) benchmarks, using only 14.1M pretraining parameters (8.5M in the point encoder), and about 6 pretraining hours on a standard single GPU. These results position CrossJEPA as a performant, memory-efficient, and fast-to-train framework for 3D representation learning via knowledge distillation. We analyze CrossJEPA intuitively, theoretically, and empirically, and extensively ablate our design choices. Code will be made available.
Read the full abstract
Computer Vision
NeuroVascU-Net: A Unified Multi-Scale and Cross-Domain Adaptive Feature Fusion U-Net for Precise 3D Segmentation of Brain Vessels in Contrast-Enhanced T1 MRI
Precise 3D segmentation of cerebral vasculature from T1-weighted contrast-enhanced (T1CE) MRI is crucial for safe neurosurgical planning. Manual delineation is time-consuming and prone to inter-observer variability, while current automated methods often trade accuracy for computational cost, limiting clinical use. We present NeuroVascU-Net, the first deep learning architecture specifically designed to segment cerebrovascular structures directly from clinically standard T1CE MRI in neuro-oncology patients, addressing a gap in prior work dominated by TOF-MRA-based approaches. NeuroVascU-Net builds on a dilated U-Net and integrates two specialized modules: a Multi-Scale Contextual Feature Fusion ($MSC^2F$) module at the bottleneck and a Cross-Domain Adaptive Feature Fusion ($CDA^2F$) module at deeper hierarchical layers. $MSC^2F$ captures both local and global information via multi-scale dilated convolutions, while $CDA^2F$ dynamically integrates domain-specific features, enhancing representation while keeping computation low. The model was trained and validated on a curated dataset of T1CE scans from 137 brain tumor biopsy patients, annotated by a board-certified functional neurosurgeon. NeuroVascU-Net achieved a Dice score of 0.8609 and precision of 0.8841, accurately segmenting both major and fine vascular structures. Notably, it requires only 12.4M parameters, significantly fewer than transformer-based models such as Swin U-NetR. This balance of accuracy and efficiency positions NeuroVascU-Net as a practical solution for computer-assisted neurosurgical planning.
Read the full abstract
Computer Vision · Cross-listed from cs.LG
Categorical Equivariant Deep Learning: Category-Equivariant Neural Networks and Universal Approximation Theorems
We develop a theory of category-equivariant neural networks (CENNs) that unifies group/groupoid-equivariant networks, poset/lattice-equivariant networks, graph and sheaf neural networks. Equivariance is formulated as naturality in a topological category with Radon measures. Formulating linear and nonlinear layers in the categorical setup, we prove the equivariant universal approximation theorem in the general setting: the class of finite-depth CENNs is dense in the space of continuous equivariant transformations. We instantiate the framework for groups/groupoids, posets/lattices, graphs and cellular sheaves, deriving universal approximation theorems for them in a systematic manner. Categorical equivariant deep learning thus allows us to expand the horizons of equivariant deep learning beyond group actions, encompassing not only geometric symmetries but also contextual and compositional symmetries.
Read the full abstract
Computer Vision
4D-VGGT: A General Foundation Model with SpatioTemporal Awareness for Dynamic Scene Geometry Estimation
We investigate a challenging task of dynamic scene geometry estimation, which requires representing both spatial and temporal features. Typically, existing methods align the two features into a unified latent space to model scene geometry. However, this unified paradigm suffers from potential mismatched representation due to the heterogeneous nature between spatial and temporal features. In this work, we propose 4D-VGGT, a general foundation model with divide-and-conquer spatiotemporal representation for dynamic scene geometry. Our model is divided into three aspects: 1) Multi-setting input. We design an adaptive visual grid that supports input sequences with arbitrary numbers of views and time steps. 2) Multi-level representation. We propose a cross-view global fusion for spatial representation and a cross-time local fusion for temporal representation. 3) Multi-task prediction. We append multiple task-specific heads to spatiotemporal representations, enabling a comprehensive visual geometry estimation for dynamic scenes. Under this unified framework, these components enhance the feature discriminability and application universality of our model for dynamic scenes. In addition, we integrate multiple geometry datasets to train our model and conduct extensive experiments to verify the effectiveness of our method across various tasks on multiple dynamic scene geometry benchmarks.