视觉-语言-动作论文趋势 2023–2026

主要结论

• 视觉-语言-动作论文从 2023 年的 257 篇增长到 2026 年的 708 篇，三年间扩大到约 2.8 倍。
• CVPR 是发视觉-语言-动作论文最多的会议，2023–2026 年共 640 篇，占四个会议总量的 39%。
• 交叉方向里涨得最快的是应用 · computer vision：同时属于两个主题的论文从 2023 年的 21 篇增至 2026 年的 135 篇。
• 在「机器人与具身」的 5 个主题里，它的 2026 年论文数排第 1；同组增长最快的人形控制是 3.8 倍，视觉-语言-动作是 2.8 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026A Dual-Domain Vision Transformer with Spectral Positional Bias
• ICML 2026FOCA: Future-Oriented Conditioning for Data-Efficient Vision-Language-Action Adaptation
• ICLR 2026Curvature-Guided Task Synergy for Skeleton based Temporal Action Segmentation
• CVPR 2026ActAvatar Temporally-Aware Precise Action Control for Talking Avatars
• NeurIPS 2025Doodle to Detect: A Goofy but Powerful Approach to Skeleton-based Hand Gesture Recognition
• ICML 2025Feasible Action Search for Bandit Linear Programs via Thompson Sampling
• ICLR 2025Action Sequence Augmentation for Action Anticipation
• CVPR 2025Active Event-based Stereo Vision
• NeurIPS 2024GAIA: Rethinking Action Quality Assessment for AI-Generated Videos
• ICML 2024Causal Action Influence Aware Counterfactual Data Augmentation
• ICLR 2024Matrix Manifold Neural Networks++
• CVPR 2024End-to-End Spatio-Temporal Action Localisation with Video Transformers
• NeurIPS 2023Activity Grammars for Temporal Action Segmentation
• ICML 2023Algorithmic Collective Action in Machine Learning
• ICLR 2023A Theoretical Understanding of Shallow Vision Transformers: Learning, Generalization, and Sample Complexity
• CVPR 20233Mformer: Multi-Order Multi-Mode Transformer for Skeletal Action Recognition
• NeurIPS 2026A Statistical Framework for Algorithmic Collective Action with Multiple Collectives
• ICML 2026From Abstraction to Instantiation: Learning Behavioral Representation for Vision-Language-Action Model
• ICLR 2026EAST: Early Action Prediction Sampling Strategy with Token Masking
• CVPR 2026DarkShake-DVS Event-based Human Action Recognition under Low-light and Shaking Camera

全部视觉-语言-动作论文

共 1,639 篇，第 10 / 33 页

451. ICML 2026Norm$\times$Direction: Restoring the Missing Query Norm in Vision Linear Attention
452. ICML 2026Off-Policy Learning in Large Action Spaces: Optimization Matters More Than Estimation
453. ICML 2026On the Adversarial Robustness of Large Vision-Language Models under Visual Token Compression
454. ICML 2026Parameters as Experts: Adapting Vision Models with Dynamic Parameter Routing
455. ICML 2026PhotoAgent: Exploratory Visual Aesthetic Planning with Large Vision Models
456. ICML 2026Pretrained Vision-Language-Action Models are Surprisingly Resistant to Forgetting in Continual Learning
457. ICML 2026RADIO1D: Elastic Representations for Condensed Vision Modeling
458. ICML 2026RealisMotion: Decomposed Human Motion Control and Video Generation in the World Space
459. ICML 2026Reinforcement Learning with Action-Triggered Observations
460. ICML 2026RePack then Refine: Efficient Diffusion Transformers with Vision Foundation Models
461. ICML 2026Respecting Modality Gap in Post-hoc Out-of-distribution Detection with Pre-trained Vision-Language Models
462. ICML 2026Rethinking Depth Pruning for Vision Transformers: A Heterogeneity-Aware Perspective
463. ICML 2026Rethinking Serialization in Linear 3D Vision: Decoupling Anisotropic Geometry from Isotropic Semantics
464. ICML 2026ReViT: Rotational-equivariant Vision Transformers for Neural PDE Solvers
465. ICML 2026REViT: Roto-reflection Equivariant Convolutional Vision Transformer
466. ICML 2026RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies
467. ICML 2026RoboOmni: Actions Are Just Another Modality for Vision-Language Models
468. ICML 2026Robustifying Vision-Language Models via Test-Time Prompt Adaptation
469. ICML 2026RoCA: Robust Cross-Domain End-to-End Autonomous Driving
470. ICML 2026SCALE: Self-uncertainty Conditioned Adaptive Looking and Execution for Vision-Language-Action Models
471. ICML 2026Scaling by Diversified Experience for Vision-Language-Action Models
472. ICML 2026See What Matters: Differentiable Grid Sample Pruning for Generalizable Vision-Language-Action Model
473. ICML 2026Seeing Realism from Simulation: Efficient Video Transfer for Vision-Language-Action Data Augmentation
474. ICML 2026Self-Captioning Multimodal Interaction Tuning: Amplifying Exploitable Redundancies for Robust Vision Language Models
475. ICML 2026Semantic Robustness Certification for Vision-Language Models
476. ICML 2026SF-Mamba: Rethinking State Space Model for Vision
477. ICML 2026SkelHCC: A Hyperbolic CLIP-Driven Cache Adaptation Framework for Skeleton-based One-Shot Action Recognition
478. ICML 2026SkillNet: Hierarchical Skill Modeling for Compositional Generalization in Vision-Language Action Models
479. ICML 2026SLAP: The Semantic Least Action Principle for Variational Video-Language Modeling
480. ICML 2026SPAR: Support-Preserving Action Rectification
481. ICML 2026Spatial Memory for Out-of-Vision Manipulation in Vision-Language-Action
482. ICML 2026SpecPrune-VLA: Accelerating Vision-Language-Action Models via Action-Aware Self-Speculative Pruning
483. ICML 2026Spectral Heat Flow for Conservative Token Condensation in Vision-Language Models
484. ICML 2026StableVLA: Towards Robust Vision-Language-Action Models without Extra Data
485. ICML 2026Tabero: Learning Gentle Manipulation with Closed-Loop Force Feedback from Vision, Touch, and Language
486. ICML 2026Temporal Difference Calibration in Sequential Tasks: Application to Vision-Language-Action Models
487. ICML 2026TGV-KV: Text-Grounded KV Eviction for Vision-Language Models
488. ICML 2026The Lie We Tell: Correcting the Euclidean Fallacy in Vision Language Action Policies via Score Matching on Tangent Space
489. ICML 2026TRAP: Hijacking VLA CoT-Reasoning via Adversarial Patches
490. ICML 2026UHR-BAT: Budget-Aware Token Compression Vision-Language model for Ultra-High-Resolution Remote Sensing
491. ICML 2026Understanding Behavior Cloning with Action Quantization
492. ICML 2026Unlocking Noise-Resistant Vision: Key Architectural Secrets for Robust Models Against Gaussian Noise
493. ICML 2026VEQ: Modality-Adaptive Quantization for MoE Vision-Language Models
494. ICML 2026VERA-V: Variational Inference Framework for Jailbreaking Vision-Language Models
495. ICML 2026Vision Language Models Cannot Reason About Physical Transformation
496. ICML 2026Vision Transformer Finetuning Benefits from Non-Smooth Components
497. ICML 2026Vision-DeepResearch: Incentivizing DeepResearch Capability in Multimodal Large Language Models
498. ICML 2026Vision-Language-Action Pretraining from Large-Scale Human Videos
499. ICML 2026ViSurf: Visual Supervised-and-Reinforcement Fine-Tuning for Large Vision-and-Language Models
500. ICML 2026VLA-Arena: An Open-Source Framework for Benchmarking Vision-Language-Action Models

常见问题

2023–2026 年四大会议一共有多少篇视觉-语言-动作论文？

共 1,639 篇，按年份分别是 2023 年 257 篇、2024 年 329 篇、2025 年 345 篇、2026 年 708 篇。

NeurIPS 2026 有多少篇视觉-语言-动作论文？

NeurIPS 2026 有 204 篇视觉-语言-动作论文，上一届是 108 篇。

ICLR 2026 有多少篇视觉-语言-动作论文？

ICLR 2026 有 113 篇视觉-语言-动作论文，上一届是 64 篇。

ICML 2026 有多少篇视觉-语言-动作论文？

ICML 2026 有 126 篇视觉-语言-动作论文，上一届是 33 篇。

CVPR 2026 有多少篇视觉-语言-动作论文？

CVPR 2026 有 265 篇视觉-语言-动作论文，上一届是 140 篇。

哪个会议的视觉-语言-动作论文最多？

CVPR，2023–2026 年共 640 篇。