CVPR 2025 accepted papers on Vision-language and multimodal
448 of the 2,871 papers accepted at CVPR 2025 (15.6%) are on Vision-language and multimodal, compared with 12.1% at CVPR 2024. They fall into 6 topics.
See CVPR 2025 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Multimodal LLMs · 154 papers
- Thinking in Space: How Multimodal Large Language Models See, Remember, and Recall Spaces (92 citations)
- Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Vision-Language Models (51 citations)
- VisionZip: Longer is Better but Not Necessary in Vision Language Models (46 citations)
- VERA: Explainable Video Anomaly Detection via Verbalized Learning of Vision-Language Models (41 citations)
- Towards Zero-Shot Anomaly Detection and Reasoning with Multimodal Large Language Models (38 citations)
- NVILA: Efficient Frontier Visual Language Models (37 citations)
- OmniDrive: A Holistic Vision-Language Dataset for Autonomous Driving with Counterfactual Reasoning (34 citations)
- EarthDial: Turning Multi-sensory Earth Observations to Interactive Dialogues (32 citations)
- RoboSpatial: Teaching Spatial Understanding to 2D and 3D Vision-Language Models for Robotics (30 citations)
- VILA-M3: Enhancing Vision-Language Models with Medical Expert Knowledge (29 citations)
- DivPrune: Diversity-based Visual Token Pruning for Large Multimodal Models (27 citations)
- FastVLM: Efficient Vision Encoding for Vision Language Models (27 citations)
and 142 more in the interactive map.
Human and embodied modeling · 105 papers
- Janus: Decoupling Visual Encoding for Unified Multimodal Understanding and Generation (44 citations)
- Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detector (29 citations)
- Human-centered Interactive Learning via MLLMs for Text-to-Image Person Re-identification (22 citations)
- TokenFlow: Unified Image Tokenizer for Multimodal Understanding and Generation (22 citations)
- MM-OR: A Large Multimodal Operating Room Dataset for Semantic Understanding of High-Intensity Surgical Environments (21 citations)
- JanusFlow: Harmonizing Autoregression and Rectified Flow for Unified Multimodal Understanding and Generation (20 citations)
- LSceneLLM: Enhancing Large 3D Scene Understanding Using Adaptive Visual Preferences (20 citations)
- RADIOv2.5: Improved Baselines for Agglomerative Vision Foundation Models (17 citations)
- Dynamic Updates for Language Adaptation in Visual-Language Tracking (16 citations)
- FOCUS: Knowledge-enhanced Adaptive Visual Compression for Few-shot Whole Slide Image Classification (16 citations)
- ResCLIP: Residual Attention for Training-free Dense Vision-language Inference (16 citations)
- Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval (14 citations)
and 93 more in the interactive map.
CLIP and zero-shot · 83 papers
- AA-CLIP: Enhancing Zero-Shot Anomaly Detection via Anomaly-Aware CLIP (69 citations)
- SAIST: Segment Any Infrared Small Target Model Guided by Contrastive Language-Image Pretraining (36 citations)
- DINOv2 Meets Text: A Unified Framework for Image- and Pixel-Level Vision-Language Alignment (32 citations)
- BiomedCoOp: Learning to Prompt for Biomedical Vision-Language Models (26 citations)
- SeeGround: See and Ground for Zero-Shot Open-Vocabulary 3D Visual Grounding (26 citations)
- SkySense-O: Towards Open-World Remote Sensing Interpretation with Vision-Centric Visual-Language Modeling (25 citations)
- MMRL: Multi-Modal Representation Learning for Vision-Language Models (22 citations)
- Bayesian Prompt Flow Learning for Zero-Shot Anomaly Detection (19 citations)
- CPath-Omni: A Unified Multimodal Foundation Model for Patch and Whole Slide Image Analysis in Computational Pathology (19 citations)
- Multimodal Autoregressive Pre-training of Large Vision Encoders (19 citations)
- LamRA: Large Multimodal Model as Your Advanced Retrieval Assistant (16 citations)
- DPC: Dual-Prompt Collaboration for Tuning Vision-Language Models (15 citations)
and 71 more in the interactive map.
Cross-modal learning · 61 papers
- DCEvo: Discriminative Cross-Dimensional Evolutionary Learning for Infrared and Visible Image Fusion (76 citations)
- One Model for ALL: Low-Level Task Interaction Is a Key to Task-Agnostic Image Fusion (57 citations)
- Every SAM Drop Counts: Embracing Semantic Priors for Multi-Modality Image Fusion and Beyond (50 citations)
- EMOE: Modality-Specific Enhanced Dynamic Emotion Experts (34 citations)
- AlignMamba: Enhancing Multimodal Mamba with Local and Global Cross-modal Alignment (18 citations)
- Multi-modal Vision Pre-training for Medical Image Analysis (17 citations)
- IDEA: Inverted Text with Cooperative Deformable Aggregation for Multi-modal Object Re-Identification (16 citations)
- Language-Guided Audio-Visual Learning for Long-Term Sports Assessment (15 citations)
- Adaptive Unimodal Regulation for Balanced Multimodal Information Acquisition (13 citations)
- Chat-based Person Retrieval via Dialogue-Refined Cross-Modal Alignment (13 citations)
- Cross-Modal Interactive Perception Network with Mamba for Lung Tumor Segmentation in PET-CT Images (12 citations)
- Align-KD: Distilling Cross-Modal Alignment Knowledge for Mobile Vision-Language Large Model Enhancement (11 citations)
and 49 more in the interactive map.
Visual question answering · 23 papers
- Augmenting Multimodal LLMs with Self-Reflective Tokens for Knowledge-based Visual Question Answering (15 citations)
- VDocRAG: Retrieval-Augmented Generation over Visually-Rich Documents (15 citations)
- Cross-modal Causal Relation Alignment for Video Question Grounding (13 citations)
- MIMO: A Medical Vision Language Model with Visual Referring Multimodal Input and Pixel Grounding Multimodal Output (10 citations)
- STING-BEE: Towards Vision-Language Model for Real-World X-ray Baggage Security Inspection (10 citations)
- Embodied Scene Understanding for Vision Language Models via MetaVQA (7 citations)
- Marten: Visual Question Answering with Mask Generation for Multi-modal Document Understanding (7 citations)
- Question-Aware Gaussian Experts for Audio-Visual Question Answering (7 citations)
- Automated Generation of Challenging Multiple-Choice Questions for Vision Language Model Evaluation (6 citations)
- JTD-UAV: MLLM-Enhanced Joint Tracking and Description Framework for Anti-UAV Systems (6 citations)
- Notes-guided MLLM Reasoning: Enhancing MLLM with Knowledge and Visual Notes for Visual Question Answering (6 citations)
- Commonsense Video Question Answering through Video-Grounded Entailment Tree Reasoning (4 citations)
and 11 more in the interactive map.
Robot action and navigation · 22 papers
- CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models (58 citations)
- SimLingo: Vision-Only Closed-Loop Autonomous Driving with Language-Action Alignment (28 citations)
- Magma: A Foundation Model for Multimodal AI Agents (26 citations)
- ShowUI: One Vision-Language-Action Model for GUI Visual Agent (24 citations)
- RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concrete (23 citations)
- Towards Long-Horizon Vision-Language Navigation: Platform, Benchmark and Method (17 citations)
- MoManipVLA: Transferring Vision-language-action Models for General Mobile Manipulation (16 citations)
- OmniManip: Towards General Robotic Manipulation via Object-Centric Interaction Primitives as Spatial Constraints (15 citations)
- Code-as-Monitor: Constraint-aware Visual Programming for Reactive and Proactive Robotic Failure Detection (13 citations)
- RoboGround: Robotic Manipulation with Grounded Vision-Language Priors (11 citations)
- Mitigating the Human-Robot Domain Discrepancy in Visual Pre-training for Robotic Manipulation (7 citations)
- Object-Centric Prompt-Driven Vision-Language-Action Model for Robotic Manipulation (7 citations)
and 10 more in the interactive map.