🧑 Human Understanding

🎞️ ECCV2026 · 119 paper notes

📌 Same area in other venues: 🧠 NeurIPS2026 (1) · 📷 CVPR2026 (151) · 🔬 ICLR2026 (45) · 🧪 ICML2026 (5) · 🤖 AAAI2026 (20) · 🧠 NeurIPS2025 (21)

🔥 Top topics: Face & Gaze ×17 · Human Pose ×13 · Diffusion Models ×12 · Avatars ×9 · Translation ×7

FlowerDance: MeanFlow for Efficient and Refined 3D Dance Generation

FlowerDance combines MeanFlow (a flow matching variant that replaces instantaneous velocity prediction with interval-average velocity) with Physical Consistency Constraints (PCC), a BiMamba backbone, and channel-level cross-modal fusion. It generates high-quality 3D dance motions in only 5–20 sampling steps. While achieving state-of-the-art (SOTA) quality on FineDance and AIST++, its inference speed (2008 FPS) far exceeds the previous best method (MatchDance at 345 FPS), leaving ample computational budget for real-time 3D rendering.

ReMoMask: Retrieval-Augmented Masked Motion Generation

ReMoMask gives motion retrieval, training masks, and generative evidence fusion a shared body structure through HBM, TSM, and SSTA, achieving FID 0.026 and R-Precision Top-1 0.566 in the main HumanML3D experiment.

Human Mesh Modeling for Anny Body

Anny converts MakeHuman's artist-designed deformation assets into a differentiable body model without learning its shape space from scans, calibrates sampling with WHO statistics, and generates 780k Anny-One images, retaining millimeter-level adult scan fitting while improving child human mesh recovery.

Forge4D: Feed-Forward 4D Human Reconstruction and Interpolation from Uncalibrated Sparse-View Videos

Forge4D decomposes reconstruction of dynamic humans and interacting objects into scale-consistent streaming 3D Gaussian prediction and bidirectional dense motion matching, then produces intermediate times through occlusion-aware fusion; it achieves 29.0378 dB PSNR on joint novel-time and novel-view evaluation on DNA-Rendering without requiring input camera calibration parameters.

VersatileMotion: A Unified Framework for Motion Synthesis and Comprehension

VersatileMotion uses the world-space causal body tokenizer VerMoVQ to formulate text, audio, and multi-person motion as unified next-token prediction, covering nine motion tasks in one model with competitive generation and understanding under the authors' evaluation protocol.

Fast Sam 3D Body: Accelerating SAM 3D Body for Real-Time Full-Body Human Mesh Recovery

Fast SAM 3D Body keeps SAM 3D Body weights frozen while restructuring body-hand dependencies and simplifying decoding, then accelerates MHR-to-SMPL conversion with a separately trained MLP, bringing reconstruction closer to interactive rates with gains that must be interpreted alongside datasets, hardware, and accuracy changes.

ECHO: Ego-centric Modeling of Human-Object Interactions

ECHO treats human motion, object trajectories, and contacts as mutually constrained diffusion variables to recover full interactions from head and wrist tracking with known object geometry, combining mixed-data training and smooth inpainting to achieve human MPJPE of 6.8±0.1 cm and object vertex error of 33.5±0.5 cm on BEHAVE.

ActionPlan: Future-Aware Streaming Motion Synthesis via Frame-Level Action Planning

ActionPlan generates temporally aligned semantic action plans before progressively synthesizing 3D human motion, combining offline generation and streaming output in one model and reducing streaming FID from MotionStreamer's 11.790 to 5.735 on the HumanML3D-272 test set.

UniMotion: A Unified Framework for Motion-Text-Vision Understanding and Generation

UniMotion treats human motion as a continuous modality with equal status to RGB within a shared LLM. By replacing discrete tokens with a Continuous Motion Aligned VAE (CMA-VAE) and a dual-path embedder, and incorporating Dual Posterior KL Alignment (DPA) to inject visual semantic priors along with Latent Reconstruction Alignment (LRA) to resolve the motion path cold-start issue, UniMotion unifies and achieves comprehensive state-of-the-art (SOTA) performance across seven understanding, generation, and editing tasks on Motion-Text-RGB tri-modalities.

CLIP-AUTT: Test-Time Personalization with Action Unit Prompting for Fine-Grained Video Emotion Recognition

CLIP-AUTT first learns expression classification from similarities between Action Unit text and videos, then selects the lowest-prediction-entropy contiguous segment in an unseen subject's video and updates only the AU prompt embeddings, improving BioVid WAR from 78.0 for the non-personalized CLIP-AU to 81.5.

DETRPose: Real-Time End-to-End Multi-Person Pose Estimation via Modified Transformer Decoder and Novel Denoising Keypoints

DETRPose ports D-FINE's real-time detection architecture to multi-person 2D pose estimation, reworking the GroupPose decoder with a keypoint-similarity (KS, i.e. COCO OKS) driven denoising keypoint strategy, a KSVF loss, and a lightweight Pose-LQE head, yielding the first end-to-end real-time transformer pose model: 73.3 AP for DETRPose-X on COCO val, and 71.2 AP / 4.66 ms for DETRPose-L on test-dev, an order of magnitude faster than GroupPose-R50 with 37% fewer parameters.

FEEL (Force-Enhanced Egocentric Learning): A Dataset for Physical Action Understanding

FEEL synchronizes a custom piezoresistive glove with Meta Aria glasses to collect roughly 22 hours and ~2M frames of real-world kitchen manipulation in first person, pairing video, audio and force; the authors then show that force is doubly useful — it yields per-frame contact labels sharper than manual annotation (training models that lead or come close to fully supervised methods on three contact benchmarks), and it serves as a self-supervised pretraining objective that improves action recognition transfer on EPIC-Kitchens, Something-Something V2, Ego-Exo4D and Meccano.

IRG-MotionLLM: Interleaving Motion Generation, Assessment and Refinement for Text-to-Motion Generation

IRG-MotionLLM places generated motion inside an interleaved generation, assessment, and refinement loop, using three-stage training to raise HumanML3D R-Precision Top-1 from 0.496 for its MotionLLM baseline to 0.535, although FID does not improve alongside alignment in the final reinforcement-learned model.

Purify then Guide: Rethinking Domain Generalization for Multimodal Face Anti-Spoofing

This paper recasts multimodal domain-generalized face anti-spoofing (FAS) as "purify then guide": MD2A, a same-domain cross-sample differential attention module, subtracts the domain- and modality-shared residual from the fused features; RS2 then softly aligns the purified RGB/IR/DEPTH features to real/spoof subspaces defined by CLIP text prompts, while a U-shaped dual-space adaptation (U-DSA) adds task capacity without destroying the pretrained geometry — lowering average HTER from 13.63% to 4.00% on WMCA/CeFA/PADISI/SURF.

UMO: Unified In-Context Learning Unlocks Motion Foundation Model Priors

UMO fine-tunes HY-Motion with three frame-level operation embeddings and input-level temporal fusion, extending a text-to-motion model to unified completion, editing, geometric control, and reaction generation while reaching a HumanML3D text-to-motion FID of 9.460 with only 0.207M additional conditioning parameters.

ClusterStyle: Modeling Intra-Style Diversity with Prototypical Clustering for Stylized Motion Generation

ClusterStyle represents each motion style with multiple global and local prototypes and conditions a pretrained motion diffusion model through a separate style-attention branch, enabling varied expressions of the same style while achieving text–motion retrieval Top-3 accuracy of 0.708 and style recognition accuracy of 78.101% on stylized generation.

FitControler: Toward Fit-Aware Virtual Try-On

FitControler removes cues about the original garment contour, predicts the body-garment layout for a target fit, and injects intermediate features into frozen try-on models, improving fit consistency and image quality on Fit4Men without predicting real garment size or physical comfort.

InfiniteDance: Scalable 3D Dance Generation Towards in-the-wild Generalization

InfiniteDance turns online dance videos into 100.69 hours of physically and foot-restored 3D motion, then trains ChoreoLLaMA with continuous embeddings, reference-dance retrieval, and cadence experts, reducing kinematic FID from Lodge's 89.52 to 30.54 on its dataset, while slow-music evaluation and long-sequence structure remain limitations of real-world generalization.

Language-Guided Transformer Tokenizer for Human Motion Generation

This paper proposes LG-Tok, which moves language guidance from the generation stage into the tokenization stage: a set of learnable latent tokens with a fully Transformer-based tokenizer/detokenizer injects text semantics into the motion representation before quantization, and a language-drop training scheme prevents shortcut learning while enabling language-free guidance decoding in motion space, cutting FID on HumanML3D from MARDM's 0.114 to 0.057.

OmniFit: Multi-modal 3D Body Fitting via Scale-agnostic Dense Landmark Prediction

OmniFit directly predicts dense, semantically fixed body landmarks from clothed-human point clouds before optimizing SMPL-X, with an optional RGB adapter and scale prediction for appearance cues and scale ambiguity, reducing 4D-DRESS point-cloud V2V from ETCH's 2.408 cm to 0.589 cm.

Silhouette-based Gait Foundation Model

FoundationGait combines multisource silhouette sequences with independent body-part processing before encoding, preserving complementary local information as gait models grow and achieving 48.0% and 64.5% self-supervised Rank-1 on Gait3D and OU-MVLP, respectively, while supporting linear probing and fine-tuning for healthcare tasks.

TripVVT: A Large-Scale Triplet Dataset and a Coarse-Mask Baseline for In-the-Wild Video Virtual Try-On

TripVVT combines 10,031 garment-transfer triplets supervised by real videos with coarse human-mask conditioning, reaching 91% Gemini-SR on a 100-case in-the-wild benchmark while exposing the cost of full-person regeneration for preserving non-target regions.

BackTranslation2.0 -- A Linguistically Motivated Metric to Assess Sign Language Production

Proposes BackTranslation2.0, a linguistically motivated text-to-sign language translation evaluation metric. Utilizing a two-phase agent framework (10 dedicated tools for evidence extraction + 4 LLM cross-comparison modules), it produces deterministic scores across four dimensions: grammatical correctness, phonological accuracy, motion fluency, and generation fidelity. It achieves a Pearson \(r=1.00\) / Spearman \(\rho=1.00\) correlation with human judgment on known corruptions and synthetic data, significantly outperforming existing backtranslation and motion similarity baselines.

CMCC-ReID: Cross-Modality Clothing-Change Person Re-Identification

The paper introduces SYSU-CMCC, a benchmark combining modality discrepancy and clothing change, and a Progressive Identity Alignment Network (PIA) that disentangles identity from clothing before bidirectional prototype alignment, achieving 57.0% Rank-1 and 46.5% mAP for visible-to-infrared retrieval.

DiffProxy: Multi-View Human Mesh Recovery via Diffusion-Generated Dense Proxies

DiffProxy generates body segmentation and UV correspondences with a diffusion model, then recovers a human mesh through hand refinement, stochastic uncertainty estimation, and SMPL-X reprojection fitting, reaching 35.1 mm MPVPE on RICH with synthetic-image training alone, without winning every metric on every dataset.

GenLCA: 3D Diffusion for Full-Body Avatars from In-the-Wild Videos

GenLCA encodes real videos into structured 3D Gaussian tokens and supervises diffusion only in visible regions, learning text- and image-controlled full-body avatars from million-scale identity data while supporting generation, editing, and facial and body animation.

HO-Flow: Generalizable Hand-Object Interaction Generation with Latent Flow Matching

HO-Flow compresses hand motion, object trajectories, and local contact geometry into temporal latents, then generates continuous motions with context-aware autoregressive flow matching, increasing GRAB Phy from LatentHOI's 96.16% to 98.25% and sample diversity SD from 0.13 to 0.31.

Infinite Gaze Generation for Videos with Autoregressive Diffusion

This paper is the first to cast video gaze prediction as raw gaze trajectory generation over arbitrarily long videos: a compression bottleneck is inserted into UNISAL so that each video frame is reduced to a compact saliency latent injected into the diffusion backbone every 5 frames, and the last 90 gaze coordinates (3 s) serve as a clean history prefix while the model autoregressively denoises the next 45 future coordinates (continuous x,y at 30 samples per second) in a sliding window, beating six image-domain gaze/scanpath models adapted to video on Levenshtein distance, DTW and Maximum Temporal Correlation on DIEM and DHF1K (Discrete Fréchet Distance excepted).

JointHOI: Jointly Generating Contact Maps Enhances Hand Object Interaction Generation

JointHOI treats contact as an inner modality of hand–object motion: a single-stage diffusion model denoises bimanual MANO poses, the object trajectory, and per-hand frame-wise contact distance maps in one shared sequence, then uses the agreement between the self-generated contact maps and the contact implied by the synthesized geometry to guide sampling — improving both action semantics (ARCTIC Top-1 0.948) and physical plausibility (interpenetration volume less than half that of Text2HOI).

Mind-to-Face: Neural-Driven Photorealistic Avatar Synthesis via EEG Decoding

Mind-to-Face is the first framework to decode non-invasive 16-channel EEG directly into dense 256×256 3D position maps (~65k vertices) that drive 3D Gaussian Splatting bound to mesh faces, rendering photorealistic facial avatars that preserve both subject identity and emotional intensity; enabling it required a purpose-built dual-modality capture system with frame-accurate synchronization between EEG and 16 multi-view high-speed cameras.

NaP-Control: Navigating Diffusion Prior for Versatile and Fast Character Control

NaP-Control freezes a diffusion action prior and action decoder, trains only a reinforcement learning policy that selects initial noise, and balances task completion with motion smoothness in physics simulation, achieving 98.4% success and 22.5 FPS on flat-ground far-goal reaching.

OmniFall: From Staged Through Synthetic to Wild, A Unified Multi-Domain Dataset for Robust Fall Detection

OmniFall unifies eight public staged fall datasets into OF-Staged by re-annotating every frame under one clinically motivated 16-class taxonomy, adds 12k controllable synthetic videos (OF-Syn) and 818 genuine-accident videos (OF-ItW), forming a ~15k-clip, 33-hour single-view (82-hour with synchronized multi-view feeds) three-domain benchmark with the first standardized staged-to-wild protocol; fine-tuned small models and zero-shot MLLMs match on fall events, but the zero-shot models systematically confuse the clinically critical fallen state with lying.

Revisiting Avatar-As-Image: High-Fidelity Registration is All You Need

AvaImg first places a plausible naked body inside clothing, progressively fits the garment surface, and converts the result into aligned UV images, achieving 2.62 mm surface Chamfer distance and 34.48 dB rendering PSNR in the main comparison, while a frozen FLUX VAE roundtrip adds only 0.76 mm geometry error.

SignBind-LLM: Multi-Stage Modality Fusion for Sign Language Translation

SignBind-LLM splits continuous sign recognition, fingerspelling recognition, and lipreading into three CTC experts pre-trained independently, uses a Gumbel-Softmax router to dispatch video segments to the right manual expert, fuses the asynchronous manual and lipreading streams through gated aggregation plus cross-attention, and finally decodes with a Flan-T5 trained only on text pairs and fully decoupled from the visual pipeline, reaching 23.1 BLEU-4 on How2Sign, 7.0 BLEU-4 on BOBSL, and 73.6% letter accuracy on ChicagoFSWild+.

SIGNET: Motion-Level Knowledge Transfer for Cross-Language Sign Language Translation

SIGNET proposes treating multiple sign language skeleton backbones pre-trained on large corpora as frozen core motion-visual experts. By utilizing hand-prior-driven attention aggregation and learnable gating fusion, it achieves cross-language motion-level knowledge transfer. With only \(1.5\text{M}\) trainable parameters, it delivers state-of-the-art (SOTA) or comparable results on four translation benchmarks and the WLASL recognition benchmark.

SignSparK: Efficient Multilingual Sign Language Production via Sparse Keyframe Learning

SignSparK treats sparse keyframes as hard anchors during training: an ultra-light segmentor (FAST) automatically mines the onset, midpoint and offset frame of every sign, and a conditional flow matching model reconstructs the full 3D signing sequence (SMPL-X upper body + MANO hands + FLAME face) under those anchors — suppressing regression-to-the-mean while cutting inference to 10 steps (~100x faster than SOKE) and setting state of the art across four sign languages (DGS, CSL, ASL, BSL).

Social-Mamba: Socially-Aware Trajectory Forecasting with State-Space Models

Social-Mamba organizes unordered neighbors into a scannable social grid and uses Cycle Mamba for temporal, egocentric, and goal-centric interactions, achieving 0.72/0.92 m ADE/FDE on NBA-Full with 1.9M parameters, although lower computation does not mean the lowest latency in every setting.

StreamTalk: Streaming Co-Speech Gesture Generation with Key-Pose Anchoring

StreamTalk turns streaming gesture generation into a clip-level loop of coarse generation, tail key-pose retrieval, and refinement, achieving FGD 0.383 for one speaker and 0.293 for all speakers on BEAT2 while using sparse destination poses to reduce minute-scale motion drift.

WebEyeTrack: Scalable Eye-Tracking for the Browser via On-Device Few-Shot Personalization

WebEyeTrack combines monocular metric head pose, the lightweight BlazeGaze eye encoder, and on-device few-shot personalization for browser eye-tracking with an initial 9-point calibration, reaching 2.32 cm gaze error on GazeCapture while prioritizing deployment efficiency and drift resistance over leading accuracy on every dataset.

ANFI: Rethinking Neighbor Feature Interaction in Person Re-ID

ANFI learns both affinity and discrepancy interactions between neighboring person images, trains their sample-wise fusion with noisy relation supervision, and achieves 88.1% mAP on CUHK03 in the same-backbone comparison, 1.2 percentage points above its strongest comparator.

Conversational Human Audio-visual Talking Dialogue Generation

CHAT turns an anonymous scenario prompt into dialogue scripts, emotional dual-track speech with brief responses, and mutually responsive face videos with smooth transitions, achieving FID 17.33 and LSE-C 6.89 against related-task baselines while showing that synthetic dialogues can improve facial reaction model pretraining.

Cross-Species Animal Re-Identification with Semantic Consistency Learning

To tackle embedding space fragmentation caused by drastic anatomical differences and complex ecological backgrounds in cross-species animal ReID, Semantic Consistency Learning (SCL) integrates Foreground–Background Decoupled Spectral Normalization (FDSNorm) for phase-preserving de-stylization and Cross-species Neighborhood Modeling (CNM) with dynamic memory for reciprocal relational alignment, establishing new state-of-the-art generalization across 11 benchmarks.

D-Rex : Diffusion Rendering for Relightable Expressive Avatars

Addressing the trade-off between relighting and expressive animation caused by explicit 3D intrinsic decomposition in full-body avatars, D-Rex decouples relighting as an image-space post-process, driving albedo-like renderings from an off-the-shelf white-light avatar and generating photorealistic, view- and temporally consistent relit avatars under arbitrary HDR environments via LoRA fine-tuning of a video diffusion model.

DiffUE: Enhancing Utility-Unlearnability Trade-off of Unlearnable Examples via Diffusion Autoencoders

DiffUE leverages a pre-trained diffusion autoencoder (DiffAE) to map defensive perturbations of unlearnable examples from conventional pixel space into a disentangled semantic latent space, utilizing bi-level minimization optimization alongside directional attribute manipulation to preserve natural image manifold structures and visual fidelity while achieving superior robustness against adversarial training, filtering, relearning attacks, and lossy compression.

Diversity-Aware View Partitioning for Scalable VGGT

Addressing the bottleneck where viewpoint redundancy in long input sequences dilutes attention and triggers GPU out-of-memory errors in visual geometry transformers, this paper presents a training-free, plug-and-play view partitioning framework that constructs balanced, high-diversity chunks via visual dissimilarity and soft pose propagation, enabling 1000-frame reconstruction on a single A100 GPU with 3.8× lower VRAM, 6.3× speedup, and superior geometric accuracy.

Dress-ED: Instruction-Guided Editing for Virtual Try-On and Try-Off

Dress-ED constructs the first large-scale benchmark dataset (146k verified quadruplets) that unifies VTON (virtual try-on), VTOFF (virtual try-off), and text instruction-guided garment editing. It proposes a unified multimodal diffusion baseline model, Dress-EM, based on MLLM + dual-path connector + DiT, which thoroughly outperforms general-purpose editing models and domain-specific VTON models on instruction-guided garment editing tasks.

DSAR: Dual-Stream Autoregressive Modeling of Temporal Cloth Dynamics for Photorealistic Animatable Avatars

Addressing the dynamic distortion and out-of-distribution breakdown caused by treating cloth deformation as a static mapping from skeletal poses, DSAR introduces a dual-stream autoregressive architecture that explicitly combines observable geometric deformation feedback with implicit memory state retrieval, synthesizing photorealistic cloth dynamics and inertia from RGB video supervision alone.

Event Stream-based Sign Language Translation: A High-Definition Benchmark Dataset and A Novel Baseline

To overcome motion blur, lighting variations, and privacy concerns in RGB sign language translation, this paper establishes Event-CSL—the first large-scale, 1280×720 HD real-world event stream benchmark (14,827 videos)—and presents EvSLT, an end-to-end gloss-free framework coupling Mamba-based clip spatial memory aggregation with graph-guided spatiotemporal fusion (GSTF).

FaceMoE: Mixture of Experts for Low-Resolution Face Recognition

FaceMoE replaces the single FFN in Transformers with multiple sparsely activated MoE experts and a Top-k router, allowing different experts to automatically specialize in different semantic regions of the face (high-frequency texture, low-frequency smooth, and landmark regions) to achieve resolution-aware feature extraction. It comprehensively outperforms the previous SOTA on three low-resolution benchmarks (BRIAR, IJB-S, and TinyFace) while suffering almost zero degradation in HR pre-training performance.

Gaze-to-text Generation: Beyond Categorical Decoding of Human Attention

Gazette is the first generative gaze-to-text decoding framework that formulates human attention inference as open-ended natural language generation, employing LLM-synthesized think-aloud transcripts across multiple observers during training to isolate goal-specific dynamics from individual scanpath idiosyncrasies.

InterCMDM: Block-Causal Diffusion for Autoregressive Human Interaction Generation

InterCMDM introduces a block-causal latent diffusion framework for autoregressive two-person interaction generation via a Dual-Stream Causal DiT, multi-task block attention masking, and latent rollout, achieving state-of-the-art realism and semantic alignment without requiring ground-truth prefix initialization.

JacobianAvatar: Temporally Consistent Semi-rigid Avatar Reconstruction from a Monocular Video

JacobianAvatar introduces hierarchical neural Jacobian fields integrated with a screened Poisson solver, signed distance-based normal regularization, and deformation-guided residual flow loss to reconstruct temporally consistent, high-fidelity animatable avatars with detailed clothing dynamics from a single monocular video.

Kirin: Animal Motion Generation from In-the-Wild Video

Kirin turns in-the-wild videos into trainable 3D quadruped motion data, uses a reference image to supply morphology missing from text, and applies generated skeletal motion to mesh animation; on AiM3D, image conditioning reduces FID from the text-only model's 11.889 to 6.248, although motion diversity metrics decrease.

LaxMotion: Rethinking Supervision Granularity for 3D Human Motion Generation

LaxMotion challenges the prevailing reliance on coordinate-level 3D supervision in text-to-motion generation by decoupling sequences into global trajectories and relative limb vectors, supervising only with single-view 2D kinematics and structural consistency constraints to synthesize diverse and high-fidelity 3D human motions.

Learning to Attract and Repel: Dual Quality Margin Learning for Face Recognition (DQM-Face)

Addressing the vulnerability of conventional magnitude-based quality estimation to identity-agnostic noise and the absence of explicit inter-class repulsion, DQM-Face fuses feature magnitude with squeeze-and-excitation semantic attention to dynamically modulate target attraction margin \(m_1\), while enforcing a curriculum-scheduled repulsion margin \(m_2\) against non-target classes to construct a wide clearance zone and deliver state-of-the-art recognition alongside top-tier intrinsic FIQA capability.

MTVDiff: Multimodal Conditional Latent Diffusion for Enhanced Thermal-to-Visible Face Translation

MTVDiff tackles geometric discontinuities, semantic attribute mismatches, and identity degradation in thermal-to-visible face synthesis by integrating complementary depth maps and textual prompts into a frozen Stable Diffusion backbone, using a dual-branch cross-attention fusion module, gated text-to-visual feature alignment, and spatial feature transformations to achieve state-of-the-art visual quality and cross-modal face verification accuracy.

Multi-scale Object-Aware Gaze Estimation via Geometric Reasoning

This work reformulates gaze target estimation from pixel-level regression to a hierarchical reasoning problem. It first establishes candidate attention objects using object-level semantic representations, constructs a field-of-view (FOV) cone geometric prior using gaze direction to constrain the search space, and finally achieves precise localization via multi-scale residual fusion. This method achieves state-of-the-art (SOTA) performance on GazeFollow, VideoAttentionTarget, ChildPlay, and GOO-Real using only 7.1M parameters.

Odoriko: A Shape-Aware Multimodal Diffusion Framework for Human Motion

Odoriko proposes the first unified multimodal human motion generation framework. By hierarchically injecting gender and SMPL shape parameters as explicit conditioning signals into the diffusion backbone, it enables the generated motion to reflect the biological morphological characteristics of the subject, while achieving or exceeding the performance of state-of-the-art specialized methods across text-to-motion, music-to-dance, and video-to-motion estimation tasks with minimal parameters.

PhysDrape: Learning Explicit Forces and Collision Constraints for Physically Realistic Garment Draping

Addressing the computational burden of traditional physics simulations and the lack of physical fidelity and severe interpenetration in deep regression, PhysDrape uses nodal forces as an intermediary to unify a force-driven GNN, a differentiable stretching solver, and a collision handler, achieving end-to-end self-supervised physically realistic 3D garment draping without ground-truth annotations.

PIAvatar: Physically Interactive Avatars via Deformation Gradient Decoupling

PIAvatar proposes an MPM-based physically interactive 3D human avatar framework. By explicitly decoupling user-defined kinematic velocity from the deformation gradient update, it eliminates unintended internal stresses generated during motion driving. It also embeds a skeletal structure to achieve real-time tracking of deformed poses through closed-form optimization, supporting bidirectional human-human and human-object physical interactions alongside non-rigid surface deformations simultaneously within a unified MPM simulation framework for the first time.

PriorPose: Reference-Guided Joint Deformation and Alignment for Category-Level Object Pose Estimation

To resolve canonical orientation memorization in prior-free methods and error cascades in serial deform-then-align prior-based pipelines, PriorPose introduces an explicit category mean shape into a geometry-aware seeded transformer to jointly predict per-point NOCS fields and prior deformation, achieving new state-of-the-art results on strict pose thresholds under dual-space shape consistency constraints.

Q-BridgeNet: A Quantization Network for Cross-Lingual Sign Language Translation

Q-BridgeNet tackles negative cross-lingual transfer and rigid temporal boundary cutting in multilingual sign language translation by introducing adaptive temporal segmentation and shared–private residual vector quantization to learn discrete Q-units, fine-tuning a multilingual LLM for many-to-many translation.

RASA: Disentangled Spatial-Motional Priors for Cross-Identity Character Animation

Addressing severe structural mismatches in cross-identity character animation, RASA introduces a hierarchical framework that injects disentangled spatial-motional priors into a Diffusion Transformer, utilizing a Spatial Prior Calibrator (SPC) at the generative onset for 2D geometric alignment and an Inherent Motional Guider (IMG) in intermediate layers for 3D anatomical consistency.

Recognizing Co-Speech Gestures in-the-Wild

Oxford VGG introduces the first large-scale in-the-wild co-speech gesture dataset GRW (~140k video clips, 155-word vocabulary) along with temporal Transformer models that leverage hand priors and broad motion context to solve semantic gesture classification, word-level recognition, and dense frame-level temporal localization.

Reconstruction-Anchored Diffusion Model for Text-to-Motion Generation

To resolve the representation gap of pre-trained text encoders lacking kinematic dynamics and error propagation during iterative denoising, RAM anchors diffusion within a motion reconstruction latent space using self-regularization and asymmetric alignment, complemented by Reconstructive Error Guidance (REG) at inference time to achieve a state-of-the-art 0.032 FID on HumanML3D in just 20 sampling steps.

RoboGesture: Real-Time Semantic-aligned Co-Speech Gestures Generation for Humanoid Interaction

RoboGesture addresses the trilemma of semantic gesture data scarcity, modality eclipse from motion inertia, and physical self-collision in humanoid co-speech generation by unifying semi-synthetic data synthesis, hierarchical acoustic-semantic alignment, anti-inertia flow matching, and real-time MPC filtering into an end-to-end policy on a 41-DoF Unitree G1 humanoid.

Rolling Shutter Relative Pose Estimation Made Practical

This paper introduces affine correspondences (ACs) into rolling shutter (RS) relative pose estimation for the first time, deriving RS-corrected affine constraints that account for point perturbation induced scanline shifts, and builds a linearized algebraic solver using 7 ACs with algebraic degree 20 that reduces RANSAC iterations by orders of magnitude while uniquely resolving the severe translational velocity coupling.

SCORE: SubDistribution-aware Collaborative Knowledge Reinforcing for Cloth-Hybrid Lifelong Person Re-Identification

SCORE addresses the catastrophic forgetting and cyclic knowledge conflicts arising from alternating cloth-consistent and cloth-changing data in lifelong person re-identification by explicitly modeling intra-identity diversity using learnable Gaussian subprototypes, coupled with collaborative distributional reinforcement and instance-level topological retention.

Self-supervised Garment Dynamics with Persistent Wrinkles

This paper proposes the first self-supervised neural network garment simulator. By leveraging dynamic rest bending energy and physically-inspired curriculum learning, the approach explicitly models the elasto-plastic deformation of fabrics. It generates natural persistent wrinkles for the first time within a self-supervised framework, outperforming existing methods across various garments, body shapes, and motions.

SICAGE: Speaker-Independent Culture-Aware Gesture Generation using TED4C-L Dataset

SICAGE formulates culture-aware co-speech gesture generation as a domain generalization problem: treating each speaker as a domain, it extracts speaker-independent cultural representations from audio-visual and textual modalities using Fishr regularization or adversarial learning. These representations then condition a real-time diffusion generator, ALaDiT, significantly improving gesture realism, diversity, beat synchronization, and cultural consistency on the self-built 106-hour, four-culture TED4C-L dataset.

SIGNER: Temporally Grounded Sign Language Generation via Time-Resolved Conditioning

SIGNER proposes a time-resolved conditioning framework. By constructing a time-resolved Gloss conditioning sequence and injecting it via Local Temporal Fusion (LTF) during diffusion denoising, it explicitly preserves the temporal grounding of downstream sign language segments. This solves the core issues of disordered sign order and inaccurate semantics in existing sign language generation methods, significantly outperforming prior SOTA on CSL-Daily and Phoenix-2014T.

SignNet-1M: Large-Scale Multilingual Sign Language Video Dataset with Downstream Benchmarks

SignNet-1M leverages 3DGS novel view synthesis, diffusion-based scene/identity editing, and post-rendering augmentation to expand 7 public sign language corpora into approximately 1 million multilingual sign language video clips (ASL/CSL/DGS). It introduces a unified evaluation protocol (Orig/Zero-shot/Trained) to expose and mitigate the robustness vulnerabilities of existing models under view, background, and identity variations.

SignRefine: Adapting Foundational Video Models for Sign Language Generation

Addressing severe articulation blur and anatomical collapse in hand and facial regions when using general video diffusion models for sign language, SignRefine introduces decoupled local spatial adapters with CoordConv positional grounding and sink-token attention masking into a frozen DiT backbone, enabling high-fidelity and comprehensible sign language video generation.

Single-Query Person-Centric Bimanual Hand-Object Interaction Detection

Addressing the ownership ambiguity and missing object semantics of conventional hand-centric HOI approaches in multi-person scenes, this paper proposes a person-centric single-query Transformer framework that couples part-aware multi-reference deformable attention with a hand-to-query relationship matrix to jointly predict human boxes, body poses, bimanual hands, and their interaction targets end-to-end.

Sound-based Multi-Person 3D Pose Estimation

The paper presents the first attempt to estimate multi-person 3D poses solely from acoustic signals, introducing SoundMHPE which leverages an Acoustic Multi-scale Encoder for time-frequency disentanglement and a Temporal Pose Decoder with per-frame multi-queries to model intra-person dynamics and inter-person acoustic interactions.

SPARC: Single-Pass Scaling for Motion Forecasting with Conformal Bayesian Last Layers

Addressing the lack of structured uncertainty in deterministic motion predictors and the high runtime overhead of Bayesian or ensemble approaches, SPARC introduces a single-pass Bayesian-conformal framework that analytically maps feature leverage to an epistemic scale, inflates graph-temporal structured Gaussian covariance without altering correlation structure, and yields 95% marginal prediction tubes with finite-sample coverage guarantees.

SyncCache: Exploiting Asymmetric Dynamics for Fast Audio-Driven Portrait Animation

SyncCache proposes a training-free feature caching acceleration scheme for DiT-based audio-driven portrait animation. By employing Spatial Asymmetric Detection (SAP) to prioritize recalculations in the facial region, Modal Decoupled Caching (MDC) to skip heavy visual modules while refreshing lightweight audio modules step-by-step, and memory-adaptive offline DP-based cache planning, it achieves 4.12x and 3.75x acceleration on HunyuanVideo-Avatar and Wan-S2V respectively, with near-lossless visual quality and lip-sync performance.

Towards Flexible, Natural, Efficient Interaction for Conversational Talking Face Generation

InterTalk establishes a motion-driven conversational talking face framework that integrates a Responsive Context Encoder, disentangled motion refinement, and an iterative generation strategy, eliminating participant count restrictions and delivering high-fidelity, real-time (30+ FPS) multi-party dialogue animations.

Towards Real-World Wearable Motion Reconstruction

Addressing the unpredictable and heterogeneous combinations of everyday wearable devices, this paper introduces a multimodal benchmark synchronizing commodity smartphones, smartwatches, pressure-sensing insoles, and a VR headset with high-precision 120-camera markerless MoCap, and proposes WHIP, a conditional flow-matching generative model that achieves robust, physically plausible full-body motion reconstruction from arbitrary sensor subsets alongside systematic quantification of sensor complementarity.

Unified and Efficient Point-Line Local Features

UPAL introduces a unified and lightweight local feature extractor that jointly extracts keypoints, line segments, and descriptors within a single ALIKED-based backbone using multi-teacher distillation and GPU-accelerated LSD, achieving a 4x speedup and 10x smaller parameter footprint while matching or surpassing SOTA accuracy.

WiFi-JEPA: Self-supervised Learning for WiFi-CSI 3D Human Pose Estimation

WiFi-JEPA preserves the physical structure of the CSI tensor as (channels C, time T, links L) via CSI-specific tokenization, masks entire Tx–Rx antenna links and lets JEPA predict the masked links' representations in latent space (rather than reconstructing hardware-contaminated raw signals), and supplements pre-training with ray-tracing simulation over randomized geometric primitives (sim-object); on PiW3D it achieves 76.8/93.5 mm single/multi-person MPJPE, beating all WiFi baselines, cuts mean cross-environment error by 48.2% versus the baseline, while SimMIM/MAE/BYOL/MoCoV3 — four vision-native SSL objectives — all degrade below training from scratch under identical backbones and protocols.

BiCE-HG: A Bi-Conditional Egocentric Hand Gesture Dataset for Intelligent Reality Systems

BiCE-HG crosses three mobility states with two lighting configurations and provides spatial illuminance records alongside hand skeleton data; its dynamic baseline reaches 89.24% validation accuracy, but the evaluation is not participant-independent and does not establish cross-lighting generalization.

CoDePose: Multi-View 3D Human Pose Estimation via Coupled 2D-3D Denoising Diffusion

Rather than treating 2D detections as immutable observations, CoDePose jointly denoises 2D and 3D poses while retaining detections as conditioning, combining cross-view fusion with stabilizing supervision to reduce MPJPE from FusionFormer's 15.1 mm to 12.1 mm on Human3.6M with ViTPose, four views, and 27 frames.

Constrained Rotation Optimization: Revisiting Crop-Based Gaze Estimation

CROp models the gaze-coordinate change associated with face cropping as a constrained approximate rotation and supplements it with joint head-pose supervision, reducing within-dataset angular error on ETH-XGaze from 6.45° to 5.55° with detector-based preprocessing, but it does not outperform normalization under every accurate-annotation or severe bounding-box-noise condition.

CtrlCoMo: Controllable Co-Speech Motion Generation with Gesture–Action Disentanglement

Addressing the interference between co-speech gestures and semantic actions, CtrlCoMo introduces Pyramid-VQ with layer-wise decaying CLIP semantic regularization to hierarchically separate low-frequency actions from high-frequency gestures, enabling text-driven action synthesis and explicit AdaLN-based gesture intensity modulation.

EgoExoMoCap: Distributed Human Motion Capture via Ego- and Exocentric Body Tracking from Head-Mounted Devices

Addressing the reliance of traditional motion capture on bulky multi-camera rigs or obtrusive suits and the failure of single egocentric wearable devices to faithfully reconstruct lower-body motions, EgoExoMoCap introduces a distributed mocap framework requiring only two or more people wearing head-mounted devices (such as Project Aria glasses); it combines continuous egocentric tracking with intermittent exocentric canonicalized 3D rays and DINOv3-based learned visibility gating to achieve robust, high-fidelity 3D global motion reconstruction even under severe occlusions and out-of-view movements.

Escaping the Low-Frequency Bias: Adversarial Frequency Perturbation for Generalisable Gaze Estimation

This paper demonstrates that cross-domain performance degradation in gaze estimation is primarily driven by overfitting to source-domain low-frequency components; it proposes an Adversarial Low-Frequency Perturbation (ALFP) framework that dynamically mixes low-frequency amplitude statistics via AdaIN using an instance-specific generator (PGNet), significantly improving cross-domain generalization without requiring any target-domain data.

Every Dog Has Its Day, Probably: A Balanced Synthetic Benchmark and Probabilistic Modeling for 3D Dog Pose Estimation

Addressing systemic viewpoint bias, restricted pose diversity, and annotation degradation in existing benchmarks, this paper builds a 4.2M-image balanced synthetic dataset with decoupled pose-shape-texture libraries and pixel-perfect 3D alignment, paired with an MMDiT Rectified Flow probabilistic framework that resolves severe self-occlusion via Mode-of-N aggregation, achieving state-of-the-art results on real-world and synthetic benchmarks trained solely on synthetic data.

FlowFace: Rectifying Identity Conditioning with Riemannian Geometry for Face Generation

FlowFace tackles identity drift and attribute order sensitivity in tuning-free face customization by proposing a fiber-bundle dual-stream manifold encoder, Lie-algebraic attribute composition via truncated BCH expansion, and training-time local SPD metric regularization, boosting identity preservation and composition stability without modifying the inference sampler.

Granular Semantic Cognition for Visible-Infrared Person Re-Identification

To address the lack of color in infrared imagery that hinders cross-modal alignment, this paper proposes Granular Semantic Cognition (GSC), which progressively refines shared semantic prototypes across stages to guide channel-wise textual color compensation into corresponding infrared local regions, regularized by a fusion-weighted identity contrastive loss to suppress cross-identity ambiguity.

HolisticSemGes: Semantic Grounding of Holistic Co-Speech Gesture Generation with Contrastive Flow-Matching

HolisticSemGes overcomes cross-articulator decoupling and rhythmic motion collapse in co-speech gesture synthesis by introducing a Semantics-Aware Composite Module and a Contrastive Flow Matching objective that explicitly repels incongruent audio-text motion trajectories.

InclusiveHuman-10K: Towards Inclusive Human Parsing Beyond the Intact-Limb Assumption

InclusiveHuman-10K (IH-10K) addresses the long-standing "intact-limb assumption" in human body parsing by establishing the first inclusive benchmark featuring 10,555 images, 14,939 person instances, and 160,133 pixel-wise masks across residual limbs, three tiers of prostheses, and mobility assistive devices alongside a subgroup-resolved evaluation protocol.

InterEdit: Navigating Text-Guided 3D Dyadic Human Motion Editing

InterEdit proposes a new task, Text-guided 3D Dyadic Motion Editing (TMME), builds the first large-scale dyadic motion editing dataset InterEdit3D (comprising 5,161 source-target-text triplets), and designs the InterEdit method based on conditional diffusion models. By utilizing semantic-aware plan token alignment to capture high-level editing intentions and interaction-aware frequency token alignment to constrain interaction rhythm using DCT frequency band energy, InterEdit significantly outperforms four baseline methods in editing faithfulness and motion realism.

K-Mask: Kinematic-Aware Masked Modeling for Controllable Text-to-Motion Synthesis

K-Mask introduces a kinematic-aware motion generation framework that couples an anatomically grounded residual VQ-VAE (KG-RVQ) with a latent-aware kinematic dropout (LAKD) loss and spatiotemporal axis-factorized masked transformers, achieving competitive generation fidelity and precise limb-level localized control without separate limb-specific encoders.

Kinematics-Agnostic 3D Human Motion Prediction via Equivariant Latent Diffusion

EquiFusion introduces the first kinematics-agnostic stochastic human motion prediction framework by explicitly incorporating the skeleton adjacency matrix as input and establishing an end-to-end permutation-equivariant latent diffusion model, bypassing skeleton-specific hard-coding and retargeting errors to achieve zero-shot generalization across unseen body topologies and occluded limbs.

MoScale: Autoregressive Next-Scale Prediction for Human Motion Generation and Editing

ScaleMoGen (MoScale) re-conceptualizes text-driven 3D human motion generation from conventional temporal-axis next-token prediction into a coarse-to-fine next-scale autoregressive prediction framework over topology-aware skeletal-temporal bitwise token maps, achieving state-of-the-art performance on HumanML3D and SnapMoGen with only 7 sampling steps and offering seamless zero-shot motion editing.

Noise-Robust Face Recognition via Non-target Similarity Distribution Guided Sample Selection

By identifying that the cosine-similarity distribution of discarded noisy targets matches that of clean non-target pairs, DDLN achieves noise-rate-agnostic sample selection with only 0.2%~0.3% computational overhead while establishing state-of-the-art recognition accuracy under extreme label noise.

Noise-Robust Facial Expression Recognition via Mamba-driven Neighbor Weight Refinement

To combat crowdsourced annotation ambiguity and label noise in in-the-wild facial expression recognition, Mamba-NWR combines cross-attention routing initialization with Mamba state space iterative refinement to dynamically balance historical states and aggregate robust neighborhood label distributions.

OrthoTailor: Geometric Orthogonalization for Conflict-Free Unified Fashion Generation

To resolve severe inter-task gradient conflict and negative transfer in shared LoRA adaptation for multi-task fashion generation, OrthoTailor introduces Orthogonal Subspace Projection (OSP) in the bottleneck and parameter-free Fisher-guided Negative Guidance (FNG) during inference, achieving state-of-the-art results across virtual try-on, garment reconstruction, and pose transfer within a single unified model.

PartCHOI: Part-Aware Guidance for Clothed Human-Object Interaction Generation

By reformulating single-view clothed human-object interaction (CHOI) as a human-centered scene where interacting objects are treated as body parts, PartCHOI introduces training-free H-GPE for closed-form pose estimation and O-GRM for Jacobian-based contact deformation, delivering physically plausible and penetration-free 3D interactive scenes.

PHOSA: Photorealistic 3D Sign Avatar Modeling and Benchmark

Addressing the lack of photorealism and fine-grained hand/facial fidelity in mesh-based sign language avatars, this paper introduces MVSign—the first 16-view Chinese sign language benchmark co-designed with Deaf experts—coupled with a hybrid multi-model SMPL-X fitting pipeline and a body-head-hand decoupled 3D Gaussian representation featuring partial hand kinematic decoupling.

Practice Makes Perfect: From Explicit Decomposition to Reinforced Latent Planning in Text-to-Human Motion

Addressing the severe inference latency of explicit Chain-of-Thought (CoT) in text-to-motion generation, LaCT-Motion mirrors human skill acquisition ("practice, internalization, perfection") by progressively compressing variable-length explicit reasoning steps into continuous latent tokens via a multi-pass curriculum, subsequently refined by latent GRPO to surpass explicit CoT generation quality with zero textual reasoning overhead at test time.

PriSM: Parsing and Style-Mixed Consistency for Unsupervised Domain Adaptation in Facial Landmark Detection

PriSM addresses the breakdown of self-training under large stylistic and geometric domain shifts by validating pseudo-labels via a dual-consistency framework—combining high-level face parsing alignment with fine-grained StyleGAN analysis-by-synthesis reconstruction—achieving state-of-the-art accuracy on CariFace and ArtiFace and remarkable zero-shot generalization across unseen domains.

Prosthesis-Aware 3D Human Pose Estimation: A Dataset and Benchmark for RSP Users

Addressing the lack of dedicated 3D pose estimation tools for running-specific prosthesis (RSP) users, this paper presents RSP3D, the first real-world motion capture dataset comprising 5.6M multi-view frames, along with a hybrid baseline that aligns model-based natural body joints with model-free prosthesis geometric reconstruction.

Reference-Free Quality Assessment for Virtual Try-On via Human Feedback

To tackle the lack of ground-truth reference images in real-world virtual try-on deployment, this paper constructs VTON-QBench—the largest curated human feedback benchmark with 62,688 try-on images and 431,800 annotations—and introduces VTON-IQA, a reference-free evaluator with Interleaved Cross-Attention that achieves near-human quality prediction and model ranking.

Revisiting the Volumetric Data of 4DME: Compression, Extension and Benchmarking for Micro-Expression Analysis

To tackle extreme volumetric data size, stereo reconstruction artifacts, and the lack of standardized benchmarks in 4D facial micro-expression action unit (ME-AU) detection, this work systematically cleans and doubles 4DME into a culturally balanced release, introduces the dual-stream VoluME network, and validates that MPEG V-PCC compression achieves an approximately 3000× data reduction with only ~1% F1 loss.

S2Gest: Split-Scan State Space Models for Dynamic Hand Gesture Recognition

Designed for continuous always-on local execution on consumer-grade hardware, S2Gest integrates a spatial-first multi-scale stem with adaptive feature modulation and introduces a parameter-neutral Split-Scan S2-Block that partitions channels to capture bidirectional temporal dynamics in linear time, matching heavy 3D CNNs and Transformers with under 3M parameters at over 100 clips/s.

SemGAN: A Semantic and Hierarchical Adversarial Network for 3D Human Pose Estimation

To tackle the lack of cross-dimensional motion semantic consistency and human kinematic hierarchy in 2D-to-3D lifting, SemGAN couples explicit coordinate regression with an implicit adversarial framework—leveraging a Semantic Prototype-guided Hierarchical Generator (SPH-Generator) and an Entropy-based Cross-Frequency Discriminator (ECF-Discriminator) to achieve a state-of-the-art 39.7mm MPJPE on Human3.6M.

SemLight: Distilled Semantic–Geometric Fusion for Efficient Local Feature Matching

Addressing matching ambiguities in low-texture, repetitive, and illumination-varying environments, SemLight distills high-level semantic priors from a vision foundation model into a compact CNN backbone and dynamically modulates appearance and surface normals via channel-wise reliability gating and 3-token cross-modal attention, delivering a 17.8% relative gain in AUC@5 over XFeat on MegaDepth while maintaining a real-time GPU runtime of 5.4 ms.

SFM: Taming State Space Models for Text-to-Motion via Spatial-Frequency Modeling

Addressing the quadratic complexity of Transformers and the restricted receptive fields of CNNs, SFM introduces a dual-axis state-space diffusion framework that decouples spatial and frequency modeling within a UNet denoiser, attaining state-of-the-art motion fidelity and linguistic alignment in linear time.

SHINE-PPG: Non-Lambertian Intrinsic Decomposition for Illumination-Robust rPPG

To overcome severe physiological signal corruption caused by environmental illumination variations and facial specular highlights in camera-based physiological monitoring, SHINE-PPG proposes a self-supervised non-Lambertian intrinsic decomposition framework that explicitly decouples facial videos into diffuse reflectance, environmental illumination, and sparse specular highlights, combined with a learnable AdaIN adversarial illumination enhancement strategy to achieve high-fidelity rPPG signal recovery under challenging out-of-distribution dynamic lighting.

SplatCtrlA: Generalizable Single Image to Fully Controllable 3D Avatar

Addressing texture degradation, structural distortion, and uncoordinated facial/hand dynamics in single-image drivable avatar creation, SplatCtrlA presents an end-to-end feed-forward framework built on SMPL-X UV Gaussian parameterization, powered by a 28.5k 3D-consistent synthetic Gaussian avatar dataset, a decoupled input-aware decoder, and a Gaussian-blending facial enhancement module to achieve second-level reconstruction of photorealistic, fully controllable 3D avatars.

Text Dictates, Music Decorates: Energy-based Attention for Editable Dance Motion Generation

STREAM proposes BEAM, a bimodal decoupled energy-based attention mechanism, which allows text to control the semantic structure of the dance motion ("what to do") while the music merely decorates its temporal rhythm ("when to do"). It resolves the modality collapse problem in joint text-music generation via mathematically guaranteed hierarchical classifier-free guidance, achieving SOTA on the newly annotated Motorica++ dataset.

TRAM: Fine-Tuning-Free Test-Time Adaptation with Few Bonafide Samples for Generalized Face Anti-Spoofing

Addressing the scarcity of target-domain spoof samples and the prohibitive cost of model fine-tuning in generalized face anti-spoofing (FAS), TRAM introduces a fine-tuning-free test-time adaptation framework that pairs incoming test images with a few target-domain bonafide anchors and performs relative image-pair-text contrastive learning with photometric prompt learning and consensus attention regularization.

Video-Text Alignment Model for Sign Language Translation

Addressing the severe temporal discrepancy between sign gesture order and spoken word order in gloss-free sign language translation, VTaMo introduces an explicit multi-granularity alignment framework that reorders visual features via entropy-regularized optimal transport with a learnable null token, establishes global geometric calibration through an orthogonal transformation, and enforces token-level grounding via position-aligned contrastive learning, establishing new state-of-the-art benchmarks across four datasets.

VIGA: View-Conditioned and Identity-Guided Adaptation for Aerial-Ground Person Re-Identification

To tackle extreme geometric deformations and overwhelming background clutter in aerial-ground person re-identification, VIGA introduces identity-guided soft masking to suppress background noise while preserving 2D spatial topology, and view-conditioned low-rank adaptation where lightweight hyper-networks dynamically modulate decoder feed-forward weights, achieving state-of-the-art cross-platform retrieval on LAGPeR and AG-ReID.v2.

Visible Yet Unrecognizable: Frequency-Selective Facial Privacy via Attention

Addressing the core privacy-utility tradeoff where existing de-identification either ruins visual quality or produces completely unfamiliar faces, MIRAGE leverages Discrete Wavelet Transform to separate face images into low- and high-frequency components, perturbing identity via AdaIN cross-attention on low frequencies while preserving visual attributes via CBAM self-attention on high frequencies, backed by a frequency-aware consistency loss to fool nine deep face recognition models while retaining realistic visual appearance.

WAPR: A Foundation Model for Wide-Angle Refinement in Unseen Object Pose Estimation

To tackle large angular deviations caused by sparse candidate poses in unseen object 6D pose estimation, WAPR introduces a wide-angle pose refinement foundation model trained on SA6D (~2M RGB-D images) with rotational symmetry prior canonicalization and an angle-balanced loss, correcting rotational misalignments up to \(\pm 90^\circ\) and achieving sub-second SOTA performance across BOP core benchmarks with only 12 candidate poses.

When Rubrics Fail: Error Enumeration as Reward for Reference-Free RL Post-Training

To resolve the failure of rubric-based rewards in reference-free tasks like virtual try-on where no single ideal output exists, this paper introduces Implicit Error Counting (IEC) with robust group calibration, achieving superior on-policy GRPO alignment over Rubrics as Rewards (RaR) and SFT at half the judge compute.