Research map / Video / Video-language understanding
VideoChat-A1: Thinking with Long Videos by Chain-of-Shot Reasoning
Zikang Wang, Boyu Chen, Zhengrong Yue, Yi Wang, Yu Qiao, Limin Wang, Yali Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add video glance pre-judging step (ablation) | VideoChat-A1 (Qwen2.5-VL-7B) · MLVU | accuracy: 70.0 -> 71.9 | Improved |
| add shot partition only (ablation) | VideoChat-A1 (Qwen2.5-VL-7B) · MLVU | accuracy: 66.2 -> 68.5 | Improved |
| add chain-of-shot only (ablation) | VideoChat-A1 (Qwen2.5-VL-7B) · MLVU | accuracy: 66.2 -> 67.0 | Improved |
| add both shot partition and chain-of-shot (ablation) | VideoChat-A1 (Qwen2.5-VL-7B) · MLVU | accuracy: 66.2 -> 71.9 | Improved |
| semantic shot partition vs cluster partition (ablation) | VideoChat-A1 (Qwen2.5-VL-7B) · LongVideoBench | accuracy: 58.4 -> 64.2 | Improved |
| semantic shot partition vs average partition (ablation) | VideoChat-A1 (Qwen2.5-VL-7B) · LongVideoBench | accuracy: 57.7 -> 64.2 | Improved |
| increase max reflection rounds from 1 to 3 (ablation) | VideoChat-A1 (Qwen2.5-VL-7B) · MLVU | accuracy: 66.2 -> 71.9 | Improved |
| increase max rounds from 3 to 4 (ablation) | VideoChat-A1 (Qwen2.5-VL-7B) · MLVU | accuracy: 71.9 -> 72.1 | No clear effect |
| 6 initial shots vs 3 or 9 (ablation) | VideoChat-A1 (Qwen2.5-VL-7B) · MLVU | accuracy: 70.8 / 71.9 / 71.8 (3/6/9) | Mixed |
| wrap InternVL2.5-8B in chain-of-shot agent | InternVL2.5-8B · Video-MME (w/ subs) | accuracy: 66.9 -> 77.0 | Improved |
| agent with fewer frames vs GPT-4o | VideoChat-A1 (InternVL2.5-8B) · frames and time | average frames: 384 -> 35.9 | Improved |
Later papers that compared with this method
1 later accepted paper reports VideoChatA1 as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Symphony: A Cognitively-Inspired Multi-Agent System for Long-Video Understanding (CVPR 2026 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Temporal Chain of Thought: Long-Video Understanding by Thinking in Frames (NeurIPS 2025 · similarity 0.90)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.89)
- TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understanding (CVPR 2024 · similarity 0.89)
- VideoReasonBench: Can MLLMs Perform Vision-Centric Complex Video Reasoning? (ICLR 2026 · similarity 0.89)
- SVBench: A Benchmark with Temporal Multi-Turn Dialogues for Streaming Video Understanding (ICLR 2025 · similarity 0.89)
- VideoChat-R1.5: Visual Test-Time Scaling to Reinforce Multimodal Reasoning by Iterative Perception (NeurIPS 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
