Saved in:
| Main Authors: | , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2501.04336 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917313057390592 |
|---|---|
| author | Huang, Zeyi Ji, Yuyang Wang, Xiaofang Mehta, Nikhil Xiao, Tong Lee, Donghyun Vanvalkenburgh, Sigmund Zha, Shengxin Lai, Bolin Ren, Yiqiu Yu, Licheng Zhang, Ning Lee, Yong Jae Liu, Miao |
| author_facet | Huang, Zeyi Ji, Yuyang Wang, Xiaofang Mehta, Nikhil Xiao, Tong Lee, Donghyun Vanvalkenburgh, Sigmund Zha, Shengxin Lai, Bolin Ren, Yiqiu Yu, Licheng Zhang, Ning Lee, Yong Jae Liu, Miao |
| contents | Long-form video understanding with Large Vision Language Models is challenged by the need to analyze temporally dispersed yet spatially concentrated key moments within limited context windows. In this work, we introduce VideoMindPalace, a new framework inspired by the "Mind Palace", which organizes critical video moments into a topologically structured semantic graph. VideoMindPalace organizes key information through (i) hand-object tracking and interaction, (ii) clustered activity zones representing specific areas of recurring activities, and (iii) environment layout mapping, allowing natural language parsing by LLMs to provide grounded insights on spatio-temporal and 3D context. In addition, we propose the Video MindPalace Benchmark (VMB), to assess human-like reasoning, including spatial localization, temporal reasoning, and layout-aware sequential understanding. Evaluated on VMB and established video QA datasets, including EgoSchema, NExT-QA, IntentQA, and the Active Memories Benchmark, VideoMindPalace demonstrates notable gains in spatio-temporal coherence and human-aligned reasoning, advancing long-form video analysis capabilities in VLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_04336 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Building a Mind Palace: Structuring Environment-Grounded Semantic Graphs for Effective Long Video Analysis with LLMs Huang, Zeyi Ji, Yuyang Wang, Xiaofang Mehta, Nikhil Xiao, Tong Lee, Donghyun Vanvalkenburgh, Sigmund Zha, Shengxin Lai, Bolin Ren, Yiqiu Yu, Licheng Zhang, Ning Lee, Yong Jae Liu, Miao Computer Vision and Pattern Recognition Long-form video understanding with Large Vision Language Models is challenged by the need to analyze temporally dispersed yet spatially concentrated key moments within limited context windows. In this work, we introduce VideoMindPalace, a new framework inspired by the "Mind Palace", which organizes critical video moments into a topologically structured semantic graph. VideoMindPalace organizes key information through (i) hand-object tracking and interaction, (ii) clustered activity zones representing specific areas of recurring activities, and (iii) environment layout mapping, allowing natural language parsing by LLMs to provide grounded insights on spatio-temporal and 3D context. In addition, we propose the Video MindPalace Benchmark (VMB), to assess human-like reasoning, including spatial localization, temporal reasoning, and layout-aware sequential understanding. Evaluated on VMB and established video QA datasets, including EgoSchema, NExT-QA, IntentQA, and the Active Memories Benchmark, VideoMindPalace demonstrates notable gains in spatio-temporal coherence and human-aligned reasoning, advancing long-form video analysis capabilities in VLMs. |
| title | Building a Mind Palace: Structuring Environment-Grounded Semantic Graphs for Effective Long Video Analysis with LLMs |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2501.04336 |