NeurIPS 2025 accepted papers on Vision-language and multimodal
539 of the 5,812 papers accepted at NeurIPS 2025 (9.3%) are on Vision-language and multimodal, compared with 7.0% at NeurIPS 2024. They fall into 6 topics.
See NeurIPS 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 · 260 papers
- MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models (119 citations)
- Spatial-MLLM: Boosting MLLM Capabilities in Visual-based Spatial Intelligence (19 citations)
- MMaDA: Multimodal Large Diffusion Language Models (16 citations)
- OCRBench v2: An Improved Benchmark for Evaluating Large Multimodal Models on Visual Text Localization and Reasoning (16 citations)
- RoboRefer: Towards Spatial Referring with Reasoning in Vision-Language Models for Robotics (9 citations)
- VITA-1.5: Towards GPT-4o Level Real-Time Vision and Speech Interaction (9 citations)
- UniTok: a Unified Tokenizer for Visual Generation and Understanding (8 citations)
- DisasterM3: A Remote Sensing Vision-Language Dataset for Disaster Damage Assessment and Response (7 citations)
- Eagle 2.5: Boosting Long-Context Post-Training for Frontier Vision-Language Models (7 citations)
- Envisioning Beyond the Pixels: Benchmarking Reasoning-Informed Visual Editing (6 citations)
- OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles (6 citations)
- Pixel Reasoner: Incentivizing Pixel Space Reasoning via Curiosity-Driven Reinforcement Learning (6 citations)
and 248 more in the interactive map.
Human and embodied modeling · 101 papers
- Show-o2: Improved Native Unified Multimodal Models (13 citations)
- V2X-Radar: A Multi-modal Dataset with 4D Radar for Cooperative Perception (7 citations)
- BioCLIP 2: Emergent Properties from Scaling Hierarchical Contrastive Learning (6 citations)
- Spot the Fake: Large Multimodal Model-Based Synthetic Image Detection with Artifact Explanation (6 citations)
- ControlFusion: A Controllable Image Fusion Network with Language-Vision Degradation Prompts (5 citations)
- CSI-Bench: A Large-Scale In-the-Wild Dataset for Multi-task WiFi Sensing (5 citations)
- OmniSVG: A Unified Scalable Vector Graphics Generation Model (5 citations)
- Scaling Computer-Use Grounding via User Interface Decomposition and Synthesis (5 citations)
- JAFAR: Jack up Any Feature at Any Resolution (4 citations)
- Vision as a Dialect: Unifying Visual Understanding and Generation via Text-Aligned Representations (4 citations)
- AlignVLM: Bridging Vision and Language Latent Spaces for Multimodal Document Understanding (3 citations)
- CPathAgent: An Agent-based Foundation Model for Interpretable High-Resolution Pathology Image Analysis Mimicking Pathologists' Diagnostic Logic (3 citations)
and 89 more in the interactive map.
Cross-modal learning · 68 papers
- A Multimodal BiMamba Network with Test-Time Adaptation for Emotion Recognition Based on Physiological Signals (2 citations)
- A TRIANGLE Enables Multimodal Alignment Beyond Cosine Similarity (2 citations)
- Collaborating Vision, Depth, and Thermal Signals for Multi-Modal Tracking: Dataset and Algorithm (2 citations)
- Generalized Contrastive Learning for Universal Multimodal Retrieval (2 citations)
- Hyper-Modality Enhancement for Multimodal Sentiment Analysis with Missing Modalities (2 citations)
- MDReID: Modality-Decoupled Learning for Any-to-Any Multi-Modal Object Re-Identification (2 citations)
- VITA-Audio: Fast Interleaved Audio-Text Token Generation for Efficient Large Speech-Language Model (2 citations)
- $\texttt{AVROBUSTBENCH}$: Benchmarking the Robustness of Audio-Visual Recognition Models at Test-Time (1 citation)
- Adaptive Gradient Masking for Balancing ID and MLLM-based Representations in Recommendation (1 citation)
- AVCD: Mitigating Hallucinations in Audio-Visual Large Language Models through Contrastive Decoding (1 citation)
- Beyond Modality Collapse: Representation Blending for Multimodal Dataset Distillation (1 citation)
- CovMatch: Cross-Covariance Guided Multimodal Dataset Distillation with Trainable Text Encoder (1 citation)
and 56 more in the interactive map.
Robot action and navigation · 49 papers
- AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-Tuning (16 citations)
- FutureSightDrive: Thinking Visually with Spatio-Temporal CoT for Autonomous Driving (13 citations)
- VLA-Cache: Efficient Vision-Language-Action Manipulation via Adaptive Token Caching (9 citations)
- Impromptu VLA: Open Weights and Open Data for Driving Vision-Language-Action Models (8 citations)
- EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models (7 citations)
- CogVLA: Cognition-Aligned Vision-Language-Action Models via Instruction-Driven Routing & Sparsification (6 citations)
- DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge (6 citations)
- ThinkAct: Vision-Language-Action Reasoning via Reinforced Visual Latent Planning (5 citations)
- Knowledge Insulating Vision-Language-Action Models: Train Fast, Run Fast, Generalize Better (4 citations)
- SoFar: Language-Grounded Orientation Bridges Spatial Reasoning and Object Manipulation (4 citations)
- UAV-Flow Colosseo: A Real-World Benchmark for Flying-on-a-Word UAV Imitation Learning (4 citations)
- BeliefMapNav: 3D Voxel-Based Belief Map for Zero-Shot Object Navigation (3 citations)
and 37 more in the interactive map.
CLIP and zero-shot · 47 papers
- Perception Encoder: The best visual embeddings are not at the output of the network (11 citations)
- Meta CLIP 2: A Worldwide Scaling Recipe (5 citations)
- SPAZER: Spatial-Semantic Progressive Reasoning Agent for Zero-shot 3D Visual Grounding (3 citations)
- Advancing Compositional Awareness in CLIP with Efficient Fine-Tuning (2 citations)
- Enhancing Compositional Reasoning in CLIP via Reconstruction and Alignment of Text Descriptions (2 citations)
- SensorLM: Learning the Language of Wearable Sensors (2 citations)
- SuperCLIP: CLIP with Simple Classification Supervision (2 citations)
- The Illusion of Progress? A Critical Look at Test-Time Adaptation for Vision-Language Models (2 citations)
- An Information-theoretical Framework for Understanding Out-of-distribution Detection with Pretrained Vision-Language Models (1 citation)
- Few-Shot Learning from Gigapixel Images via Hierarchical Vision-Language Alignment and Modeling (1 citation)
- Learning a Cross-Modal Schrödinger Bridge for Visual Domain Generalization (1 citation)
- MAPLE: Multi-scale Attribute-enhanced Prompt Learning for Few-shot Whole Slide Image Classification (1 citation)
and 35 more in the interactive map.
Visual question answering · 14 papers
- Benchmarking Retrieval-Augmented Multimomal Generation for Document Question Answering (3 citations)
- Knowledge-based Visual Question Answer with Multimodal Processing, Retrieval and Filtering (3 citations)
- Robo2VLM: Improving Visual Question Answering using Large-Scale Robot Manipulation Data (2 citations)
- CoralVQA: A Large-Scale Visual Question Answering Dataset for Coral Reef Image Understanding (1 citation)
- IndEgo: A Dataset of Industrial Scenarios and Collaborative Work for Egocentric Assistants (1 citation)
- Vision-and-Language Training Helps Deploy Taxonomic Knowledge but Does Not Fundamentally Alter It (1 citation)
- Escaping the SpuriVerse: Can Large Vision-Language Models Generalize Beyond Seen Spurious Correlations?
- FOCUS: Internal MLLM Representations for Efficient Fine-Grained Visual Question Answering
- Language‑Bias‑Resilient Visual Question Answering via Adaptive Multi‑Margin Collaborative Debiasing
- mmWalk: Towards Multi-modal Multi-view Walking Assistance
- SMMILE: An expert-driven benchmark for multimodal medical in-context learning
- TCM-Ladder: A Benchmark for Multimodal Question Answering on Traditional Chinese Medicine
and 2 more in the interactive map.