AAAI 2026 Papers

Blind image quality assessment (BIQA) methods often incorporate auxiliary tasks to improve performance. However, existing approaches face limitations due to insufficient integration and a lack of flexible uncertainty estimation, leading to suboptimal performance. To address these challenges, we propose a multitasks-based Deep Evidential Fusion Network (DEFNet) for BIQA, which performs multitask optimization with the assistance of scene and distortion type classification tasks. To achieve a more robust and reliable representation, we design a novel trustworthy information fusion strategy. It first combines diverse features and patterns across sub-regions to enhance information richness, and then performs local-global information fusion by balancing fine-grained details with coarse-grained context. Moreover, DEFNet exploits advanced uncertainty estimation technique inspired by evidential learning with the help of normal-inverse gamma distribution mixture. Extensive experiments on both synthetic and authentic distortion datasets demonstrate the effectiveness and robustness of the proposed framework. Additional evaluation and analysis are carried out to highlight its strong generalization capability and adaptability to previously unseen scenarios.

Diffusion-based talking head models generate high-quality, photorealistic videos but suffer from slow inference, limiting practical applications. Existing acceleration methods for gen- eral diffusion models fail to exploit the temporal and spatial redundancies unique to talking head generation. In this paper, we propose a task-specific framework addressing these inefficiencies through two key innovations. First, we introduce Lightning-fast Caching-based Parallel denoising prediction (LightningCP), caching static features to bypass most model layers in inference time. We also enable parallel prediction using cached features and estimated noisy latents as inputs, efficiently bypassing sequential sampling. Second, we propose Decoupled Foreground Attention (DFA) to further accelerate attention computations, exploiting the spatial decoupling in talking head videos to restrict attention to dynamic foreground regions. Additionally, we remove reference features in certain layers to bring extra speedup. Extensive experiments demonstrate that our framework significantly improves inference speed while preserving video quality.

Few-shot action recognition (FSAR) has recently made notable progress through set matching and efficient adaptation of large-scale pre-trained models. However, two key limitations persist. First, existing set matching metrics typically rely on cosine similarity to measure inter-frame linear dependencies and then perform matching with only instance-level information, thus failing to capture more complex patterns such as nonlinear relationships and overlooking task-specific cues. Second, for efficient adaptation of CLIP to FSAR, recent work performing fine-tuning via skip-fusion layers (which we refer to as side layers) has significantly reduced memory cost. However, the newly introduced side layers are often difficult to optimize under limited data conditions. To address these limitations, we propose TS-FSAR, a framework comprising three components: (1) a visual Ladder Side Network (LSN) for efficient CLIP fine-tuning; (2) a metric called Task-Specific Distance Correlation Matching (TS-DCM), which uses alpha-distance correlation to model both linear and nonlinear inter-frame dependencies and leverages a task prototype to enable task-specific matching; and (3) a Guiding LSN with Adapted CLIP (GLAC) module, which regularizes LSN using the adapted frozen CLIP to improve training for better α-distance correlation estimation under limited supervision. Extensive experiments on five widely-used benchmarks demonstrate that our TS-FSAR yields superior performance compared to prior state-of-the-arts.

Traditionally, AI research in medical diagnosis has largely centered on image analysis. While this has led to notable advancements, the absence of patient-reported symptoms continues to hinder diagnostic accuracy. To address this, we propose a Pre-Consultation Dialogue Framework (PCDF) that mimics real-world diagnostic procedures, where doctors iteratively query patients before reaching a conclusion. Specifically, we simulate diagnostic dialogues between two vision–language models (VLMs): a DocVLM, which generates follow-up questions based on the image and dialogue history, and a PatientVLM, which responds using a symptom profile derived from the ground-truth diagnosis. We additionally conducted a small-scale clinical validation of the synthetic symptoms generated by our framework, with licensed clinicians confirming their clinical relevance, symptom coverage, and overall realism. These findings indicate that the resulting DocVLM–PatientVLM interactions form coherent, multi-turn consultations paired with images and diagnoses, which we then use to fine-tune the DocVLM. This dialogue-based supervision leads to substantial gains over image-only training, highlighting the value of realistic symptom elicitation for diagnosis.

Atmospheric turbulence (AT) severely degrades video quality by introducing distortions such as geometric warping, blur, and temporal flickering, posing significant challenges to both visual clarity and temporal consistency. Current state-of-the-art methods are based on transformer, 3D architectures and require multi-frame input, but their large computational cost and memory usage limit real-time deployment. In this work, we propose RMFAT, Recurrent Multi-scale Feature Atmospheric Turbulence Mitigator, designed for efficient and temporally consistent video restoration under AT conditions. RMFAT adopts a lightweight recurrent framework that restores each frame using only two inputs at a time, significantly reducing temporal window size and computational burden. It further integrates multi-scale feature encoding and decoding with temporal warping modules at both encoder and decoder stages to enhance spatial detail and temporal coherence. Extensive experiments conducted on synthetic and real-world atmospheric turbulence datasets demonstrate that RMFAT not only outperforms existing methods in terms of clarity restoration (with nearly a 9% improvement in SSIM) but also achieves significantly improved inference speed (achieving a more than fourfold reduction), making it particularly suitable for real-time atmospheric turbulence suppression tasks.

Modeling large-scale landscapes is a foundational yet time-consuming task in many 3D applications, typically requiring substantial expertise. Recently, Text-to-3D techniques have emerged as a promising, beginner-friendly prototyping approach for generating 3D content from textual input. However, existing methods either produce unusable, problematic geometries, or fail to fully capture the user's complex intent from the input text—making it difficult to generate high-quality landscape assets with controllable spatial and geographic features. In this paper, we present LandCraft, a novel AI-assisted authoring tool that enables the rapid creation of high-quality landscape scenes based on user descriptions. Our system employs a coarse-to-fine generation process: Initially, large language and deep generative models concretize textual ideas into abstract representations that capture essential landscape features, such as spatial and geographic characteristics. Then, we leverage a comprehensive procedural generation module to synthesize the detailed, structurally consistent 3D landscapes based on these inferred representations. LandCraft can effectively generate production-ready 3D scene assets that can be seamlessly exported to external game engines or modeling software, enabling immediate practical use.

Diffusion models have recently been adopted for point cloud upsampling due to their effectiveness in solving ill-posed problems. However, existing upsampling methods often struggle with inefficiencies, as they generate dense point clouds by mapping Gaussian noise to data, overlooking the geometric information already present in sparse inputs. To address this, we propose PUFM, a novel Point Cloud Upsampling via Flow Matching, which learns to directly transform sparse point clouds into their high-fidelity dense counterparts. Our approach first applies midpoint interpolation to densify the sparse input. Then, we construct a continuous interpolant between sparse and dense point clouds and train a neural network to estimate the velocity field for flow matching. Given the unordered nature of point clouds, we introduce a pre-alignment step based on Earth Mover's Distance (EMD) optimization to ensure coherent and meaningful interpolation between sparse and dense representations. This results in a more stable and efficient learning trajectory during flow matching. Experiments on synthetic benchmarks demonstrate that our method delivers superior upsampling quality but with fewer sampling steps. Further experiments on ScanNet and KITTI also show that our approach generalizes well to real-world RGB-D and LiDAR point clouds, making it more practical for real-world applications.

Recent advancements in video diffusion models have significantly enhanced audio-driven portrait animation. However, current methods still suffer from flickering, identity drift, and poor audio-visual synchronization. These issues primarily stem from entangled appearance-motion representations and unstable inference strategies. In this paper, we introduce ConsistTalk, a novel intensity-controllable and temporally consistent talking head generation framework with diffusion noise search inference. First, we propose an optical flow-guided temporal module (OFT) that decouples motion features from static appearance by leveraging facial optical flow, thereby reducing visual flicker and improving temporal consistency. Second, we present an Audio-to-Intensity (A2I) model obtained through multimodal teacher-student knowledge distillation. By transforming audio and facial velocity features into a frame-wise intensity sequence, the A2I model enables joint modeling of audio and visual motion, resulting in more natural dynamics. This further enables fine-grained, frame-wise control of motion dynamics while maintaining tight audio-visual synchronization. Third, we introduce a diffusion noise initialization strategy (IC-Init). By enforcing explicit constraints on background coherence and motion continuity during inference-time noise search, we achieve better identity preservation and refine motion dynamics compared to the current autoregressive strategy. Extensive experiments demonstrate that ConsistTalk significantly outperforms prior methods in reducing flicker, preserving identity, and delivering temporally stable, high-fidelity talking head videos.

Existing gaze estimation models often struggle to generalize to unseen users, primarily due to significant variations in individual appearance. Empirical observations reveal that performance improves when the visual appearance of test subjects closely resembles that of training subjects. Motivated by this, we propose a generalizable gaze estimation framework MoEGaze based on the Mixture of Experts (MoE) architecture. During training, the model extracts appearance features from facial images and uses them to route samples to specialized gaze expert networks, each tailored to a specific subset of appearances. Rather than directly predicting gaze, each expert outputs intermediate gaze features, which are dynamically aggregated according to the input appearance and then mapped to gaze prediction. This dynamic routing design enables the model to effectively adapt to users with diverse appearances, while also facilitating easier training on sub-datasets with smaller appearance variations. Extensive experiments demonstrate that our method achieves superior cross-domain performance compared to existing approaches, with an average improvement of 27.6% across four cross-domain metrics over the baseline. Furthermore, MoEGaze surpasses baselines trained on the full dataset while requiring only 10% of the training data.

RGB-to-RAW reconstruction, or the reverse modeling of a camera Image Signal Processing (ISP) pipeline, aims to recover high-fidelity RAW data from RGB images. Despite notable progress, existing learning-based methods typically treat this task as a direct regression objective and still struggle with detail inconsistency and color deviation, due to the ill-posed nature of inverse ISP and the inherent information loss in quantized RGB images. To address these limitations, we pioneer a generative perspective by reformulating RGB-to-RAW reconstruction as a deterministic latent transport problem and introduce a novel framework named RAW-Flow, which leverages flow matching to learn a deterministic vector field in latent space, to effectively bridge the gap between RGB and RAW representations and enable accurate reconstruction of structural details and color information. To further enhance latent transport, we introduce a cross-scale context guidance module that injects hierarchical RGB features into the flow estimation process. Moreover, we design a Dual-domain Latent Autoencoder (DLAE) with a feature alignment constraint to support the proposed latent transport framework, which jointly encodes RGB and RAW inputs while promoting stable training and high-fidelity reconstruction. Extensive experiments demonstrate that RAW-Flow outperforms state-of-the-art approaches both quantitatively and visually.

Precise event spotting (PES) aims to recognize fine-grained events at exact moments and has become a key component of sports analytics. This task is particularly challenging due to rapid succession, motion blur, and subtle visual differences. Consequently, most existing methods rely on domain-specific, end-to-end training with large labeled datasets and often struggle in few-shot conditions due to their dependence on pixel- or pose-based inputs alone. However, obtaining large labeled datasets is practically hard. We propose a Unified Multi-Entity Graph Network (UMEG-Net) for few-shot PES. UMEG-Net integrates human skeletons and sport-specific object keypoints into a unified graph and features an efficient spatio-temporal extraction module based on advanced GCN and multi-scale temporal shift. To further enhance performance, we employ multimodal distillation to transfer knowledge from keypoint-based graphs to visual representations. Our approach achieves robust performance with limited labeled data and significantly outperforms baseline models in few-shot settings, providing a scalable and effective solution for few-shot PES.

Automated radiology report generation (R2Gen) has advanced significantly, yet evaluation remains challenging due to the complexity of assessing report quality. Traditional metrics often misalign with human judgments, failing to identify specific deficiencies. To address this, we introduce ReFINE, a framework for training an Evaluation Model using a novel margin-based reward enforcement loss. This approach decomposes report quality into fine-grained sub-scores across user-defined criteria, improving interpretability. Leveraging GPT-4, we generate diverse training data with paired accepted and rejected reports to train our model under a reward-based system. The trained ReFINE Score provides both granular sub-scores and an aggregated quality assessment, enabling criterion-specific evaluation. Experimental results demonstrate ReFINE's superior alignment with human judgments, outperforming traditional metrics in model selection. Its robustness is validated across three expert-annotated datasets—including chest X-rays and multimodal reports covering 9 imaging modalities—and under two distinct scoring systems.

Nighttime flares, caused by complex scattering and reflections from artificial light sources, significantly degrade image quality and hinder downstream visual tasks. Existing deflare networks usually struggle to jointly capture and fuse latent spatial and frequency features. In this paper, we propose a novel Wavelet-guided and Gated-enhanced Spatial-frequency Fusion Network (WGSF-Net) for nighttime flare removal. WGSF-Net is primarily composed of two key modules: Wavelet-guided Fusion Block (WFB) and Local-Global Block (LGB). Specifically, WFB integrates a Multi-level Wavelet Enhancement Block (MWEB) and a Spatial-Frequency Fusion Network (SFFN) to effectively extract hierarchical spatial and frequency features through a coarse-to-fine strategy based on multi-level wavelet decomposition. To better suppress flare artifacts, LGB is designed to jointly capture local and global information: a Gated-Enhanced Attention Block (GEAB) selectively amplifies critical local features through a gated network and a difference network, and the subsequent SFFN performs global spatial-frequency fusion via depthwise separable convolution and partial Fourier convolution. This design enables LGB to effectively disentangle flare-corrupted regions and restore fine-grained details, making it particularly suited for challenging real-world flare scenarios. Extensive experiments on both synthetic and real datasets show that WGSF-Net achieves state-of-the-art performance in nighttime flare removal, outperforming existing methods across five evaluation metrics.

Restoring nighttime images affected by multiple adverse weather conditions is a practical yet under-explored research problem, as multiple weather degradations usually coexist in the real world alongside various lighting effects at night. This paper first explores the challenging multi-weather nighttime image restoration task, where various types of weather degradations are intertwined with flare effects. To support the research, we contribute the AllWeatherNight dataset, featuring large-scale nighttime images with diverse compositional degradations. By employing illumination-aware degradation generation, our dataset significantly enhances the realism of synthetic degradations in nighttime scenes, providing a more reliable benchmark for model training and evaluation. Additionally, we propose ClearNight, a unified nighttime image restoration framework, which effectively removes complex degradations in one go. Specifically, ClearNight extracts Retinex-based dual priors and explicitly guides the network to focus on uneven illumination regions and intrinsic texture contents respectively, thereby enhancing restoration effectiveness in nighttime scenarios. Moreover, to more effectively model the common and unique characteristics of multiple weather degradations, ClearNight performs weather-aware dynamic specificity and commonality collaboration that adaptively allocates optimal sub-networks associated with specific weather types. Comprehensive experiments on both synthetic and real-world images demonstrate the necessity of the AllWeatherNight dataset and the superior performance of ClearNight.

3D visual grounding (3DVG) identifies objects in 3D scenes from language descriptions. Existing zero-shot approaches leverage 2D vision–language models (VLMs) by converting 3D spatial information (SI) into forms amenable to VLM processing, typically as composite inputs such as specified-view renderings or video sequences with overlaid object markers. However, this VLM ⊕ SI paradigm yields entangled visual representations that compel the VLM to process entire cluttered cues, making it hard to exploit spatial–semantic relationships effectively. In this work, we propose a new VLM ⊗ SI paradigm that externalizes the 3D SI into a form enabling the VLM to incrementally retrieve only what it needs during reasoning. We instantiate this paradigm with a novel View-on-Graph (VoG) method, which organizes the scene into a multi-modal, multi-layer scene graph and allows the VLM to operate as an active agent that selectively accesses necessary cues as it traverses the scene. This design offers two intrinsic advantages: (i) by structuring 3D context into a spatially and semantically coherent scene graph rather than confounding the VLM with densely entangled visual inputs, it lowers the VLM's reasoning difficulty; and (ii) by actively exploring and reasoning over the scene graph, it naturally produces transparent, step-by-step traces for interpretable 3DVG. Extensive experiments show that VoG achieves state-of-the-art zero-shot performance, establishing structured scene exploration as a promising strategy for advancing zero-shot 3DVG.

Controllable generation of realistic LiDAR scenes is crucial for applications such as autonomous driving and robotics. While recent diffusion-based models achieve high-fidelity LiDAR generation, they lack explicit control over foreground objects and spatial relationships, limiting their usefulness for scenario simulation and safety validation. To address these limitations, we propose Large-scale Layout-guided LiDAR generation model ("La La LiDAR"), a novel layout-guided generative framework that introduces semantic-enhanced scene graph diffusion with relation-aware contextual conditioning for structured LiDAR layout generation, followed by foreground-aware control injection for complete scene generation. This enables customizable control over object placement while ensuring spatial and semantic consistency. To support our structured LiDAR generation, we introduce Waymo-SG and nuScenes-SG, two large-scale LiDAR scene graph datasets, along with new evaluation metrics for layout synthesis. Extensive experiments demonstrate that La La LiDAR achieves state-of-the-art performance in both LiDAR generation and downstream perception tasks, establishing a new benchmark for controllable 3D scene generation.

Fracture injuries often lead to complex bone fragmentations, posing significant challenges for accurate segmentation in surgical planning and trauma assessment. Manual annotation of each fragment is time-consuming and inconsistent, while existing automated methods often fail to separate individual fragments due to the wide variation in fracture types, irregular fracture surface, and close inter-fragment contact. To address these challenges, we introduce FracSegmentator, a deep learning approach for bone fragment instance segmentation. The model takes extracted bone regions in CT as input and isolates individual fragments by identifying fracture surfaces and separating closely contacting structures. Central to our approach is a Trauma-Prior-Guided Contrastive Learning module, which incorporates clinical knowledge through memory-based attention to better distinguish fractured surfaces from healthy regions. We evaluate FracSegmentator on four datasets that cover a range of anatomical sites and fracture patterns. The method achieves state-of-the-art results across all datasets and demonstrates strong generalization capabilities. By delivering accurate and efficient fragment-level segmentation, FracSegmentator supports critical downstream tasks such as automated fracture diagnosis, surgical planning, and preoperative reduction simulation.

Multi-modal object Re-IDentification (ReID) is devoted to retrieving specific objects through the exploitation of complementary multi-modal image information. Existing methods mainly concentrate on the fusion of multi-modal features, yet neglecting the background interference. Besides, current multi-modal fusion methods often focus on aligning modality pairs but suffer from multi-modal consistency alignment. To address these issues, we propose a novel selective interaction and global-local alignment framework called Signal for multi-modal object ReID. Specifically, we first propose a Selective Interaction Module (SIM) to select important patch tokens with intra-modal and inter-modal information. These important patch tokens engage in the interaction with class tokens, thereby yielding more discriminative features. Then, we propose a Global Alignment Module (GAM) to simultaneously align multi-modal features by minimizing the volume of 3D polyhedra in the gramian space. Meanwhile, we propose a Local Alignment Module (LAM) to align local features in a shift-aware manner. With these modules, our proposed framework could extract more discriminative features for object ReID. Extensive experiments on three multi-modal object ReID benchmarks (i.e., RGBNT201, RGBNT100, MSVR310) validate the effectiveness of our method.

Large vision-language models (LVLMs) excel at visual understanding but face efficiency challenges due to quadratic complexity when processing long multimodal contexts. While token compression can reduce computational costs, existing approaches are designed for single-view LVLMs and fail to account for the unique multi-view characteristics of high-resolution LVLMs that use dynamic cropping. Current methods treat all tokens uniformly, yet our analysis shows that global thumbnails can naturally guide the compression of local crops by providing holistic context for evaluating informativeness. In this paper, we first analyze the dynamic cropping strategy, revealing both the complementary relationship between thumbnails and crops and the distinct characteristics across different crops. Based on these insights, we propose ''Global Compression Commander'' (GlobalCom2), a novel plug-and-play token compression framework for high-resolution LVLMs. GlobalCom2 uses the thumbnail as a ''commander'' to adaptively guide the compression of local crops, preserving informative details while removing redundancy. Extensive experiments demonstrate that GlobalCom2 maintains over 90% of model performance while compressing 90% of visual tokens, reducing FLOPs to 9.1% and peak memory usage to 60% of the original.