👋 About Me

I am currently pursuing my Ph.D. at the Institute of Computing Technology , Chinese Academy of Sciences , advised by Prof. Zhaoqi Wang. Concurrently, I serve as a Research Intern at AMAP , Alibaba , where I work closely with Xiangxiang Chu. I am deeply grateful for the opportunity to collaborate with exceptional researchers including Prof. Shuo Li, Prof. Yujun Cai, and Prof. Yiwei Wang. Their mentorship and insights have profoundly shaped my academic journey.

My research interest includes Vision-Language Model (VLM), Large Language Model (LLM), Embodied Agents, Multimodal AI, and 3D Vision. I have published 20+ papers at the top international AI conferences such as NeurIPS, ICLR, ICML, CVPR, ICCV, AAAI, etc.

📚 Research Interests

• Foundation Models & Pre-training 🔥🔥
  • Vision-Language Models (VLMs) / Vision-Language Action (VLA) / Spatial Intelligence
• Model Enhancement & Post-training 🔥🔥
  • Reasoning & Alignment / Tool-Augmented RL / NLP-Enhanced Training
• Model Interpretation 🔥🔥
  • Mechanistic Interpretability / Factuality, Truthfulness, and Social Good
• Real-World Applications🔥🔥
  • Embodied Agents / AI for Science / Biomedical Engineering

🔥 Main News

• 2026.02:  🎉🎉 Our work ADE-CoT has been Accepted by CVPR 2026.
• 2026.01:  🎉🎉 Our work Video-STAR has been Accepted by ICLR 2026.
• 2026.01:  🎉🎉 Our work AutoDrive-R² has been Accepted by ICLR 2026.
• 2025.11:  🎉🎉 We propose ADE-CoT, which is now available on ArXiv!
• 2025.11:  🎉🎉 We propose Reasoning-VLA, which is now available on ArXiv!
• 2025.10:  🎉🎉 Our work DVP-MVS++ has been Accepted by TCSVT 2025.
• 2025.10:  🎉🎉 We propose Video-STAR, which is now available on ArXiv!
• 2025.08:  🎉🎉 Our work AutoDrive-R² was reported by AutoDrive Heart (自动驾驶之心)
• 2025.08:  🎉🎉 We propose AutoDrive-R², which is now available on ArXiv!
• 2025.06:  🎉🎉 We propose DVP-MVS++, which is now available on ArXiv!
• 2025.05:  🎉🎉 Our work SED-MVS has been Accepted by TCSVT 2025.
• 2024.12:  🎉🎉 We propose SED-MVS, which is now available on ArXiv!
• 2024.12:  🎉🎉 Our work DVP-MVS has been Accepted by AAAI 2025.
• 2024.12:  🎉🎉 Our work MSP-MVS has been Accepted by AAAI 2025.
• 2024.08:  🎉🎉 We propose DVP-MVS, which is now available on ArXiv!
• 2024.08:  🎉🎉 We propose MSP-MVS, which is now available on ArXiv!
• 2024.05:  🎉🎉 Our work TSAR-MVS has been Accepted by PR 2024.
• 2024.01:  🎉🎉 We propose TSAR-MVS, which is now available on ArXiv!
• 2023.12:  🎉🎉 Our work SD-MVS has been Accepted by AAAI 2024.
• 2023.09:  🎉🎉 We propose SD-MVS, which is now available on ArXiv!

📝 Main Publications

Multimodal LLMs Post-Training

ICLR

Video-STAR: Reinforcing Zero-shot Video Understanding with Tools

Think with Videos Tool-Using Agent Multi-turn Agentic RL

Yuan Z., Qu X., Qian, C., Chen, R., Tang, J., Sun L., Chu X., Zhang D., Wang Y., Cai Y., Li S.

[Paper]

ICLR

AutoDrive-R²: Incentivizing Reasoning and Self-Reflection Capacity for VLA Model in Autonomous Driving

Multimodal Reasoning Autonomous Driving Open-World Applications

Featured by AutoDrive Heart (自动驾驶之心)

Yuan Z., Tang, J., Luo, J., Chen, R., Qian, C., Sun, L., Cai Y., Zhang D., Li, S

[Paper]

Preprint

Reasoning-VLA: A Fast and General Vision-Language-Action Reasoning Model for Autonomous Driving

Multimodal Reasoning Autonomous Driving Open-World Applications

Zhang D.*, Yuan Z.*, Chen Z., Liao C., Chen Y., Shen F., Zhou Q., Chua T.

[Paper]

Generative Foundation Model

Preprint

From Scale to Speed: Adaptive Test-Time Scaling for Image Editing

Generation Model Image Editing Text-to-Image Generation

Qu X.*, Yuan Z.*, Tang J., Chen R., Tang D., Yu M., Sun L., Bai Y., Chu X., Gou G,., Xiong G., Cai Y.

Preprint

Recovering Degradations with Generative Model: A Consistency-aware Distillation Network for Infrared and Visible Image Fusion

Generation Model Image Editing Text-to-Image Generation

Yu H.*, Yuan Z.*, Bai Y., Li J., Liu J., Li S., Sun L., Chu X.

3D Vision

TCSVT

DVP-MVS++: Synergize Depth-Normal-Edge and Harmonized Visibility Prior for Multi-View Stereo

Yuan Z., Zhang, D., Li, Z., Qian, C., Chen, J., Chen, Y., Chen K., Mao T., Li Z, Jiang H., Wang, Z

IEEE Transactions on Circuits and Systems for Video Technology (IEEE TCSVT) (Under Review), 2025.

[Paper]

TCSVT

SED-MVS: Segmentation-Driven and Edge-Aligned Deformation Multi-View Stereo with Depth Restoration and Occlusion Constraint

Yuan Z.., Yang, Z., Cai, Y., Wu, K., Liu, M., Zhang, D., Jiang H, Li Z., Wang, Z.

IEEE Transactions on Circuits and Systems for Video Technology (IEEE TCSVT), 2025.

[Paper]

AAAI

DVP-MVS: Synergize Depth-Edge and Visibility Prior for Multi-View Stereo

Yuan Z.., Luo, J., Shen, F., Li, Z., Liu, C., Mao, T., Wang, Z.

AAAI Conference on Artificial Intelligence (AAAI), 2025.

[Paper] [Code]

AAAI

MSP-MVS: Multi-granularity segmentation prior guided multi-view stereo

Yuan Z., Liu, C., Shen, F., Li, Z., Luo, J., Mao, T., Wang, Z.

AAAI Conference on Artificial Intelligence (AAAI), 2025.

[Paper] [Code]

AAAI

SD-MVS: Segmentation-driven deformation multi-view stereo with spherical refinement and em optimization

Yuan Z., Cao, J., Li, Z., Jiang, H., Wang, Z.

AAAI Conference on Artificial Intelligence (AAAI), 2024.

[Paper] [Code]

PR

TSAR-MVS: Textureless-aware segmentation and correlative refinement guided multi-view stereo

Yuan Z., Cao, J., Wang, Z., Li, Z..

Pattern Recognition (PR), 2024.

[Paper] [Code]

📖 All Publications

• ICLR 2025 Video-STAR: Reinforcing Zero-shot Video Understanding with Tools. Z Yuan, X Qu, C Qian, et al.
• ICLR 2025 AutoDrive-R2: Incentivizing Reasoning and Self-Reflection Capacity for VLA Model in Autonomous Driving. Z Yuan, J Tang, J Luo, et al.
• Preprint What if Agents Could Imagine? Reinforcing Open-Vocabulary HOI Comprehension through Generation. Z Yuan, X Qu, J Tang, et al.
• Preprint From Scale to Speed: Adaptive Test-Time Scaling for Image Editing. X Qu, Z Yuan, et al.
• Preprint Recovering Degradations with Generative Model: A Consistency-aware Distillation Network for Infrared and Visible Image Fusion. H Yu, Z Yuan, Y Bai, et al.
• Preprint Pure Vision Language Action (VLA) Models: A Comprehensive Survey. D Zhang, J Sun, C Hu, X Wu, Z Yuan, et al.
• Preprint AT-Drive: Exploiting Adversarial Transfer for End-to-end Autonomous Driving. D Zhang, Z Yuan, K Huang, et al.
• Preprint ADDI: A Simplified E2E Autonomous Driving Model with Distinct Experts and Implicit Interactions. D Zhang, Z Yuan, Chen Y., et al.
• IEEE JBHI 2025 EMPOWER: Evolutionary Medical Prompt Optimization With Reinforcement Learning. Y Chen, Y He, J Yang, D Zhang, Z Yuan, et al.
• IEEE TCSVT 2025 DVP-MVS++: Synergize Depth-Normal-Edge and Harmonized Visibility Prior for Multi-View Stereo. Z Yuan, D Zhang, Z Li, et al.
• NIPS 2025 InstructHOI: Context-Aware Instruction for Multi-Modal Reasoning in Human-Object Interaction Detection. J Luo, W Ren , Q Zheng, Y Zhang, Z Yuan, et al.
• IEEE TCSVT 2025 Learning multi-view stereo with geometry-aware prior. K Chen, Z Yuan, H Xiao, T Mao, et al.
• HCII 2025 MR-IntelliAssist: A World Cognition Agent Enabling Adaptive Human-AI Symbiosis in Industry 4.0., C Liu, Z Yuan, Y Wang, Y Yin, et al.
• IEEE TCSVT 2025 SED-MVS: Segmentation-Driven and Edge-Aligned Deformation Multi-View Stereo with Depth Restoration and Occlusion Constraint., Z Yuan, Z Yang, Y Cai, et al.
• AAAI 2025 Dual-level precision edges guided multi-view stereo with accurate planarization., K Chen, Z Yuan, T Mao, et al.
• AAAI 2025 Mapexpert: Online hd map construction with simple and efficient sparse map element expert., D Zhang, D Chen, P Zhi, Y Chen, Z Yuan, et al.
• AAAI 2025 DVP-MVS: Synergize depth-edge and visibility prior for multi-view stereo., Z Yuan, J Luo, F Shen, et al.
• AAAI 2025 MSP-MVS: Multi-granularity segmentation prior guided multi-view stereo., Z Yuan, C Liu, F Shen, et al.
• IEEE TCSVT 2025 Light4gs: Lightweight compact 4d gaussian splatting generation via context model., M Liu, Q Yang, H Huang, W Huang, Z Yuan, et al.
• Preprint Adaptive label correction for robust medical image segmentation with noisy labels., C Qian, K Han, J Ding, L Liu, C Lyu, Z Yuan, et al.
• Preprint Dyncim: Dynamic curriculum for imbalanced multimodal learning., C Qian, K Han, J Wang, Z Yuan, et al.
• PR 2025 Nerf-based polarimetric multi-view stereo., J Cao, Z Yuan, T Mao, et al.
• PR 2024 Tsar-mvs: Textureless-aware segmentation and correlative refinement guided multi-view stereo., Z Yuan, J Cao, Z Wang, et al.
• AAAI 2024 Sd-mvs: Segmentation-driven deformation multi-view stereo with spherical refinement and em optimization., Z Yuan, J Cao, Z Li, et al.

🏆 Service

• Conference Reviewers: NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, AAAI
• Journal Reviewers: IJCV, TIP, TMM, TNNLS, TCSVT, PR