Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2510.05034 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909922189377536 |
|---|---|
| author | Tang, Yolo Y. Bi, Jing Liu, Pinxin Pan, Zhenyu Tan, Zhangyun Shen, Qianxiang Liu, Jiani Hua, Hang Guo, Junjia Xiao, Yunzhong Huang, Chao Wang, Zhiyuan Liang, Susan Liu, Xinyi Song, Yizhi Huang, Junhua Zhong, Jia-Xing Li, Bozheng Qi, Daiqing Zeng, Ziyun Vosoughi, Ali Song, Luchuan Zhang, Zeliang Shimada, Daiki Liu, Han Luo, Jiebo Xu, Chenliang |
| author_facet | Tang, Yolo Y. Bi, Jing Liu, Pinxin Pan, Zhenyu Tan, Zhangyun Shen, Qianxiang Liu, Jiani Hua, Hang Guo, Junjia Xiao, Yunzhong Huang, Chao Wang, Zhiyuan Liang, Susan Liu, Xinyi Song, Yizhi Huang, Junhua Zhong, Jia-Xing Li, Bozheng Qi, Daiqing Zeng, Ziyun Vosoughi, Ali Song, Luchuan Zhang, Zeliang Shimada, Daiki Liu, Han Luo, Jiebo Xu, Chenliang |
| contents | Video understanding represents the most challenging frontier in computer vision, requiring models to reason about complex spatiotemporal relationships, long-term dependencies, and multimodal evidence. The recent emergence of Video-Large Multimodal Models (Video-LMMs), which integrate visual encoders with powerful decoder-based language models, has demonstrated remarkable capabilities in video understanding tasks. However, the critical phase that transforms these models from basic perception systems into sophisticated reasoning engines, post-training, remains fragmented across the literature. This survey provides the first comprehensive examination of post-training methodologies for Video-LMMs, encompassing three fundamental pillars: supervised fine-tuning (SFT) with chain-of-thought, reinforcement learning (RL) from verifiable objectives, and test-time scaling (TTS) through enhanced inference computation. We present a structured taxonomy that clarifies the roles, interconnections, and video-specific adaptations of these techniques, addressing unique challenges such as temporal localization, spatiotemporal grounding, long video efficiency, and multimodal evidence integration. Through systematic analysis of representative methods, we synthesize key design principles, insights, and evaluation protocols while identifying critical open challenges in reward design, scalability, and cost-performance optimization. We further curate essential benchmarks, datasets, and metrics to facilitate rigorous assessment of post-training effectiveness. This survey aims to provide researchers and practitioners with a unified framework for advancing Video-LMM capabilities. Additional resources and updates are maintained at: https://github.com/yunlong10/Awesome-Video-LMM-Post-Training |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_05034 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Video-LMM Post-Training: A Deep Dive into Video Reasoning with Large Multimodal Models Tang, Yolo Y. Bi, Jing Liu, Pinxin Pan, Zhenyu Tan, Zhangyun Shen, Qianxiang Liu, Jiani Hua, Hang Guo, Junjia Xiao, Yunzhong Huang, Chao Wang, Zhiyuan Liang, Susan Liu, Xinyi Song, Yizhi Huang, Junhua Zhong, Jia-Xing Li, Bozheng Qi, Daiqing Zeng, Ziyun Vosoughi, Ali Song, Luchuan Zhang, Zeliang Shimada, Daiki Liu, Han Luo, Jiebo Xu, Chenliang Computer Vision and Pattern Recognition Video understanding represents the most challenging frontier in computer vision, requiring models to reason about complex spatiotemporal relationships, long-term dependencies, and multimodal evidence. The recent emergence of Video-Large Multimodal Models (Video-LMMs), which integrate visual encoders with powerful decoder-based language models, has demonstrated remarkable capabilities in video understanding tasks. However, the critical phase that transforms these models from basic perception systems into sophisticated reasoning engines, post-training, remains fragmented across the literature. This survey provides the first comprehensive examination of post-training methodologies for Video-LMMs, encompassing three fundamental pillars: supervised fine-tuning (SFT) with chain-of-thought, reinforcement learning (RL) from verifiable objectives, and test-time scaling (TTS) through enhanced inference computation. We present a structured taxonomy that clarifies the roles, interconnections, and video-specific adaptations of these techniques, addressing unique challenges such as temporal localization, spatiotemporal grounding, long video efficiency, and multimodal evidence integration. Through systematic analysis of representative methods, we synthesize key design principles, insights, and evaluation protocols while identifying critical open challenges in reward design, scalability, and cost-performance optimization. We further curate essential benchmarks, datasets, and metrics to facilitate rigorous assessment of post-training effectiveness. This survey aims to provide researchers and practitioners with a unified framework for advancing Video-LMM capabilities. Additional resources and updates are maintained at: https://github.com/yunlong10/Awesome-Video-LMM-Post-Training |
| title | Video-LMM Post-Training: A Deep Dive into Video Reasoning with Large Multimodal Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2510.05034 |