WACV 2026 Papers

Low-light image enhancement remains a challenging task, particularly in preserving object edge continuity and fine structural details under extreme illumination degradation. In this paper, we propose a novel model, DRWKV (Detailed Receptance Weighted Key Value), which integrates our proposed Global Edge Retinex (GER) theory, enabling effective decoupling of illumination and edge structures for enhanced edge fidelity. Secondly, we introduce Evolving WKV Attention, a spiral-scanning mechanism that captures spatial edge continuity and models irregular structures more effectively. Thirdly, we design the Bilateral Spectrum Aligner (Bi-SAB) and a tailored MS2-Loss to jointly align luminance and chrominance features, improving visual naturalness and mitigating artifacts. Extensive experiments on five LLIE benchmarks demonstrate that DRWKV achieves leading performance in PSNR, SSIM, and NIQE while maintaining low computational complexity. Furthermore, DRWKV enhances downstream performance in low-light multi-object tracking tasks, validating its generalization capabilities.

We present HAQAGen, a unified generative model for resolution-invariant NIR-to-RGB colorization that balances chromatic realism with structural fidelity. The proposed model introduces (i) a combined loss term aligning the global color statistics through differentiable histogram matching, perceptual image quality measure, and feature-based similarity to preserve texture information, (ii) local hue-saturation priors injected via Spatially Adaptive Denormalization (SPADE) to stabilize chromatic reconstruction, and (iii) texture-aware supervision within a Mamba backbone to preserve fine details. We introduce an adaptive-resolution inference engine that further enables high-resolution translation without sacrificing quality. Our proposed NIR-to-RGB translation model simultaneously enforces global color statistics and local chromatic consistency, while scaling to native resolutions without compromising texture fidelity or generalization. Extensive evaluations on FANVID, OMSIV, VCIP2020, and RGB2NIR using different evaluation metrics demonstrate consistent improvements over state-of-the-art baseline methods. HAQAGen produces images with sharper textures and natural colors, attaining significant gains as per perceptual metrics. These results position HAQAGen as a scalable and effective solution for NIR-to-RGB translation across diverse imaging scenarios.

We present SurfDist, a convolutional neural network architecture for three-dimensional volumetric instance segmentation. SurfDist is a modification of the popular model architecture StarDist-3D which enables learning instance boundaries as closed piecewise compositions of smooth parametric surfaces. This parameterization breaks StarDist-3D's coupling of instance dimension and instance voxel resolution, and it produces predictions which may be upsampled to arbitrarily high resolutions without introduction of voxelization artifacts. For datasets with blob-shaped instances, common in biomedical imaging, SurfDist can achieve higher segmentation accuracy than StarDist-3D with more compact instance parameterizations.

We introduce state invariance - robustness to changes in an object's structural form (as for a folded umbrella or for crumpled clothing) - complementing other common invariances to learn object representations for recognition and retrieval tasks. For this, we present ObjectsWithStateChange (OWSC), a novel dataset that captures variations in object appearance arising from state changes, along with pose, viewpoint, and illumination variations, to advance research in fine-grained 3D object recognition and retrieval. A key challenge is that objects within and across categories may look visually similar under certain state changes, making discrimination difficult. To address this, we propose a curriculum learning based mining strategy that progressively samples harder object pairs based on inter-object distances in the learned embedding space after each epoch, gradually sampling harder-to-distinguish examples of visually similar objects from within and across categories during training. Our ablation shows that this curriculum learning strategy enhances the model's ability to learn discriminative invariant features for fine-grained tasks, improving object recognition accuracy by 7.9% and retrieval mAP by 9.2% over prior methods on our new OWSC dataset and three other multi-view datasets, such as ModelNet40, ObjectPI, FG3D. Our OWSC dataset is available at https://github.com/sarkar-rohan/ObjectsWithStateChange .

State Space Models (SSMs) with selective scan (Mamba) have been adapted into efficient vision models. Mamba, unlike Vision Transformers, achieves linear complexity for token interactions through a recurrent hidden state process. This sequential processing is enhanced by a parallel scan algorithm, which reduces the computational time of recurrent steps from L sequential steps to log(L) parallel steps with respect to the number of input tokens (L). In this work, we propose Fast Vision Mamba (FastVim), that further reduces the computational time of the SSM block by reducing the number of recurrent steps in Vision Mamba models while still retaining model performance. By alternately pooling tokens along image dimensions across Mamba blocks, we obtain a 2x reduction in the number of parallel steps in SSM block. Our model offers up to 72.5% speedup in inference speed compared to baseline Vision Mamba models on high resolution (2048x2048) images. Our experiments demonstrate state-of-the-art performance with dramatically improved throughput in a range of tasks such as image classification, cell perturbation prediction, segmentation, and object detection. Code is made available at https://github.com/insitro/FastVim

Recent advancements in diffusion models have significantly improved the realism and generalizability of character-driven animation, enabling the synthesis of high-quality motion from just a single RGB image and a set of driving poses. Nevertheless, generating temporally coherent long-form content remains challenging. Existing approaches are constrained by computational and memory limitations, as they are typically trained on short video segments, thus performing effectively only over limited frame lengths and hindering their potential for extended coherent generation. To address these constraints, we propose TalkingPose, a novel diffusion-based framework specifically designed for producing long-form, temporally consistent human upper-body animations. TalkingPose leverages driving frames to precisely capture expressive facial and hand movements, transferring these seamlessly to a target actor through a stable diffusion backbone. To ensure continuous motion and enhance temporal coherence, we introduce a feedback-driven mechanism built upon image-based diffusion models. Notably, this mechanism does not incur additional computational costs or require secondary training stages, enabling the generation of animations with unlimited duration. Additionally, we introduce a comprehensive, large-scale dataset to serve as a new benchmark for human upper-body animation. Project page: https://dfki-av.github.io/TalkingPose

We present brat (brain report alignment transformer), a multi-view representation learning framework for brain magnetic resonance imaging (MRI) trained on MRIs paired with clinical reports. Brain MRIs present unique challenges due to the presence of numerous, highly varied, and often subtle abnormalities that are localized to a few slices within a 3D volume. To address these challenges, we introduce a brain MRI dataset 10x larger than existing ones, containing approximately 80,000 3D scans with corresponding radiology reports, and propose a multi-view pre-training approach inspired by advances in document retrieval. We develop an implicit query-feature matching mechanism and adopt concepts from quality-diversity to obtain multi-view embeddings of MRIs that are aligned with the clinical features given by report sentences. We evaluate our approach across multiple vision-language and vision tasks, demonstrating substantial performance improvements. The brat foundation models are publicly released.

Drones are increasingly used in civilian and defense domains, but reliable detection remains challenging due to their small size, fast motion, and diverse environments. Existing datasets, such as synthetic benchmarks, fail to capture real-world variability. We introduce RealDroneVision, a unified contribution that advances both dataset and methodology. First, we curate a large-scale real-world drone detection dataset comprising 173,023 images, constructed via a semi-automatic pipeline inspired by self-annotated labeling from videos, enhanced with a human-in-the-loop to iteratively reduce false positives and false negatives. This approach yields high-quality annotations with reduced manual effort. Second, we propose the Nano Object Vision Attention (NOVA) module, a drop-in replacement for YOLOv8's C2f block. By combining depthwise separable convolutions, scale-aware dilated branches, lightweight mixing, and coordinate-aware attention, our design improves small-object detection while remaining computationally efficient. Extensive benchmarks against YOLOv8m/l and YOLOv9c/e demonstrate that YOLOv8-NOVA dominates across precision (0.912), recall (0.870), and mAP@50 (0.920) while being significantly more lightweight (2.3M params, 5 MB weights). These results establish RealDroneVision as a strong foundation for advancing real-world drone detection research.

Object state changes in video reveal critical cues about human and agent activity. However, existing methods are limited to temporal localization of when the object is in its initial state (e.g., cheese block) versus when it has completed a state change (e.g., grated cheese), offering no insight into where the change is unfolding. We propose to deepen the problem by introducing the spatially-progressing object state change segmentation task. The goal is to segment at the pixel-level those regions of an object that are actionable and those that are transformed. We show that state-of-the-art VLMs and video segmentation methods struggle at this task, underscoring its difficulty and novelty. As an initial baseline, we design a VLM-based pseudo-labeling approach, state-change dynamics constraints, and a novel WhereToChange benchmark built on in-the-wild Internet videos. Experiments on two datasets validate both the challenge of the new task as well as the promise of our model for localizing exactly where and how fast objects are changing in video. We further demonstrate useful implications for tracking activity progress to benefit robotic agents. Overall, our work positions spatial OSC segmentation as a new frontier task for video understanding: one that challenges current SOTA methods and invites the community to build more robust, state-change-sensitive representations.

6D pose estimation aims at determining the object pose that best explains the camera observation. The unique solution for non-ambiguous objects can turn into a multi-modal pose distribution for symmetrical objects or when occlusions of symmetry-breaking elements happen, depending on the viewpoint.Currently, 6D pose estimation methods are benchmarked on datasets that consider, for their ground truth annotations, visual ambiguities as only related to global object symmetries, whereas they should be defined per-image to account for the camera viewpoint. We thus first propose an automatic method to re-annotate those datasets with a 6D pose distribution specific to each image, taking into account the object surface visibility in the image to correctly determine the visual ambiguities. Second, given this improved ground truth, we re-evaluate the state-of-the-art single pose methods and show that this greatly modifies the ranking of these methods. Third, as some recent works focus on estimating the complete set of solutions, we derive a precision/recall formulation to evaluate them against our image-wise distribution ground truth, making it the first benchmark for pose distribution methods on real images.

The ability to explain complex information from chart images is vital for effective data-driven decision-making. In this work, we address the challenge of generating detailed explanations alongside answering questions about charts. We present ChartQA-X, a comprehensive dataset comprising 30,799 chart samples across four chart types, each paired with contextually relevant questions, answers, and explanations. Explanations are generated and selected based on metrics such as faithfulness, informativeness, coherence, and perplexity. Our human evaluation with 245 participants shows that model-generated explanations in ChartQA-X surpass human-written explanations in accuracy and logic and are comparable in terms of clarity and overall quality. Moreover, models fine-tuned on ChartQA-X show substantial improvements across various metrics, including absolute gains of up to 24.57 points in explanation quality, 18.96 percentage points in question-answering accuracy, and 14.75 percentage points on unseen benchmarks for the same task. By integrating explanatory narratives with answers, our approach enables agents to convey complex visual information more effectively, improving comprehension and greater trust in the generated responses.

Adapting person re-identification (reID) models to new target environments remains a challenging problem that is typically addressed using unsupervised domain adaptation (UDA) methods. Recent works show that when labeled data originates from several distinct sources (for example, datasets and cameras), considering each source separately and applying multi-source domain adaptation (MSDA) typically yields higher accuracy and robustness compared to blending the sources and performing conventional UDA. However, state-of-the-art MSDA methods learn domain-specific backbone models or require access to source domain data during adaptation, resulting in significant growth in training parameters and computational cost. In this paper, a Source-free Adaptive Gated Experts (SAGE-reID) method is introduced for person reID. SAGE-reID is a cost-effective, source-free MSDA method that first trains individual source-specific low-rank adapters (LoRA) through source-free UDA. Next, a lightweight gating network is introduced and trained to dynamically assign optimal merging weights for the fusion of LoRA experts, enabling effective cross-domain knowledge transfer. While the number of backbone parameters remains constant across source domains, LoRA experts scale linearly but remain negligible in size (less than or equal to 2 percent per source), reducing both memory consumption and the risk of overfitting. Extensive experiments conducted on three challenging benchmarks -- Market-1501, DukeMTMC-reID, and MSMT17 -- indicate that SAGE-reID outperforms state-of-the-art methods while being computationally efficient.

Recent text-to-image diffusion models generate high-quality images but struggle to learn new styles, which limits the personalized content creation. In response, style-driven generation has become a popular task, wherein users supply reference images capturing the target style, complemented by text prompts that specify stylistic cues. Fine-tuning is a common approach, yet it often blindly utilizes pre-training configurations without modification, especially for noise schedules defined in terms of signal-to-noise ratio (SNR), which determines the amount of image information available at each denoising step. We discover that stylistic features predominantly emerge at low SNR range, leading current fine-tuning methods using regular noise schedules to exhibit suboptimal style alignment. We propose the Style-friendly SNR sampler, which focuses the fine-tuning on low SNR range where stylistic features emerge. We demonstrate improved generation of novel styles that cannot be described solely with a text prompt, enabling high-fidelity personalized content creation.

Comprehending long videos remains a significant challenge for Large Multi-modal Models (LMMs). Current LMMs struggle to process even minutes to hours videos due to their lack of explicit memory and retrieval mechanisms. To address this limitation, we propose RAVU (Retrieval Augmented Video Understanding), a novel framework for video understanding enhanced by retrieval with compositional reasoning over a spatio-temporal graph. We construct a graph representation of the video, capturing both spatial and temporal relationships between entities. This graph serves as a long-term memory, allowing us to track objects and their actions across time. To answer complex queries, we decompose the queries into a sequence of reasoning steps and execute these steps on the graph, retrieving relevant key information. Our approach enables more accurate understanding of long videos, particularly for queries that require multi-hop reasoning and tracking objects across frames. Our approach demonstrate superior performances with limited retrieved frames (5-10) compared with other SOTA methods and baselines on two major video QA datasets, NExT-QA and EgoSchema.

Recent breakthroughs in radiance fields, particularly 3D Gaussian Splatting (3DGS), have unlocked real-time, high-quality rendering of complex environments, enabling a wide range of applications. However, the stringent requirements of mixed reality (MR) rendering, such as rapid refresh rates, high-resolution stereo viewing, and constrained computing budgets, remain out of reach for current 3DGS techniques. Nevertheless, the wide field-of-view design of MR displays, which mimics human vision, presents a unique opportunity to exploit human visual system's own perceptual limitations to reduce computational overhead while not compromising user-perceived rendering quality. To this end, we propose a perception-guided, continuous level-of-detail (LOD) framework for 3DGS that maximizes perceived quality under given compute resources. We distill a visual quality metric, which encodes the spatial, temporal, and peripheral characteristics of human visual perception, into a lightweight, gaze-contingent model that predicts and adaptively modulates rendering LOD across the visual field based on each region's contributions to perceptual quality. This budget-driven LOD modulation, guided by both scene content and gaze behavior, enables significant computation reduction with minimal loss in perceived quality. To support low-power, untethered MR setups, we design an edge-cloud collaborative rendering framework to partially offload computation to the cloud, further reducing overhead on the edge MR devices. Objective metrics and MR user study evidence that, compared to vanilla and foveated LOD baselines, our method achieves superior trade-offs between computational efficiency and user-perceived visual quality.

Gait analysis is crucial for the diagnosis and monitoring of movement disorders like Parkinson's Disease. While computer vision models have shown potential for objectively evaluating parkinsonian gait, their effectiveness is limited by scarce clinical datasets and the challenge of collecting large and well-labelled data, impacting model accuracy and risk of bias.To address these gaps, we propose GAITGen, a novel framework that generates realistic gait sequences conditioned on specified pathology severity levels. GAITGen employs a Conditional Residual Vector Quantized Variational Autoencoder to learn disentangled representations of motion dynamics and pathology-specific factors, coupled with Mask and Residual Transformers for conditioned sequence generation. GAITGen generates realistic, diverse gait sequences across severity levels, enriching datasets and enabling large-scale model training in parkinsonian gait analysis. Experiments on our new PD-GaM (real) dataset demonstrate that GAITGen outperforms adapted state-of-the-art models in both reconstruction fidelity and generation quality, accurately capturing critical pathology-specific gait features. A clinical user study confirms the realism and clinical relevance of our generated sequences. Moreover, incorporating GAITGen-generated data into downstream tasks improves parkinsonian gait severity estimation, highlighting its potential for advancing clinical gait analysis.

Clinical photographs play a vital role in conversational computer-aided diagnosis, particularly in dermatology. However, existing skin disease benchmarks contain limitations like insufficient dataset size, the sole presence of categorical labels, the lack of expert inspections, and limited diversity in annotations. To address these shortcomings, we introduce DermEVAL, a large-scale benchmark specifically designed to evaluate the performance of Multimodal Large Language Models (MLLMs) in dermatology. Our benchmark includes image-text pairs depicting 16 distinct skin diseases, featuring a total of 11,347 representative images drawn from various dermatological datasets, carefully selected and annotated with the guidance of dermatologists. DermEVAL enables two primary tasks: visual question answering (VQA) and medical report generation (MRG), designed to simulate real-world medical diagnostics. We evaluate the performance of MLLMs in dermatology using multiple metrics, including traditional metrics and GPT-4V-based assessments. Our results indicate that accurately diagnosing skin diseases remains challenging for state-of-the-art MLLMs. We also demonstrate that fine-tuning MLLMs using DermEVAL significantly improves their performance on dermatology-related image-text tasks.

Moire patterns, caused by frequency aliasing between fine repetitive structures and a camera sensor's sampling process, have been a significant obstacle in various real-world applications, such as consumer photography and industrial defect inspection. With the advancements in deep learning algorithms, numerous studies-predominantly based on convolutional neural networks-have suggested various solutions to address this issue. Despite these efforts, existing approaches still struggle to effectively eliminate artifacts due to the diverse scales, orientations, and color shifts of moire patterns, primarily because the constrained receptive field of CNN-based architectures limits their ability to capture the complex characteristics of moire patterns. In this paper, we propose MZNet, a U-shaped network designed to bring images closer to a 'Moire-Zero' state by effectively removing moire patterns. It integrates three specialized components: Multi-Scale Dual Attention Block (MSDAB) for extracting and refining multi-scale features, Multi-Shape Large Kernel Convolution Block (MSLKB) for capturing diverse moire structures, and Feature Fusion-Based Skip Connection for enhancing information flow. Together, these components enhance local texture restoration and large-scale artifact suppression. Experiments on benchmark datasets demonstrate that MZNet achieves state-of-the-art performance on high-resolution datasets and delivers competitive results on lower-resolution dataset, while maintaining a low computational cost, suggesting that it is an efficient and practical solution for real-world applications.

Multiple Object Tracking (MOT) in dynamic and densely populated scenes presents significant challenges due to frequent occlusions, erratic object motion, and identity switches. While recent Transformer-based approaches have successfully leveraged global attention for object detection, most rely on temporal reasoning at the decoder level, leaving encoder-stage modeling underexplored. In this work, we propose an encoder-level temporal reasoning Transformer framework that embeds historical object trajectory information into the encoder stage via a tracklet memory. The encoder module, enhanced by Attention-by-Tracking, enriches visual features with temporal priors, while the decoder leverages Tracking-by-Attention to guide identity association using learned tracklet representations. To further improve temporal consistency and object localization, we introduce a gating-based temporal feature fusion mechanism that adaptively integrates multi-frame features based on cosine similarity. Our method is model-agnostic and can be applied to various Transformer-based MOT frameworks. When integrated into existing models such as TransTrack, MeMOTR, and MOTIP, it yields consistent performance improvements. Extensive experiments on DanceTrack and SportsMOT benchmarks demonstrate that our approach achieves superior tracking performance, including a HOTA score of 76.4 on SportsMOT. These results validate the effectiveness of encoder-level temporal integration and adaptive feature fusion for robust multi-object tracking in real-world scenarios.