AI Paper Atlas

Research map / Video / Video-language understanding

Chain-of-Glimpse: Search-Guided Progressive Object-Grounded Reasoning for Video Understanding

Zhixuan Wu, Quanxing Zha, Teng Wang, Genbao Xu, Wenyuan Gu, Wei Rao, Nan Ma, Bo Cheng, Soujanya Poria

ICML 2026 · Video › Video-language understanding · approach: Long video understanding · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
add object-grounded reasoning with MCTS, SFT and GRPO (full method) to Qwen2.5-VL-7BQwen2.5-VL-7B · NExTQAaccuracy (%): 79.7 -> 83.3Improved
add object-grounded reasoning with random object selection (ablation)Qwen2.5-VL-7B · NExTQAaccuracy (%): 79.7 -> 79.7No clear effect
use greedy object exploration instead of random (ablation)Qwen2.5-VL-7B · NExTQAaccuracy (%): 79.7 -> 80.9Improved
add plain CoT reasoning (ablation)Qwen2.5-VL-7B · Video-Holmesaccuracy (%): 38.8 -> 38.6No clear effect
add MCTS-distilled SFT (ablation)Qwen2.5-VL-7B · NExTQAaccuracy (%): 79.7 -> 80.6Improved
add GRPO after SFT without grounding (ablation)SFT model · NExTQAaccuracy (%): 80.6 -> 81.4Improved
add MCTS-based OGVRF to SFT (ablation)SFT model · NExTQAaccuracy (%): 80.6 -> 82.3Improved
use evidence-aware reward instead of vanilla GRPO on OGVRF (ablation)Chain-of-Glimpse · NExTQAaccuracy (%): 82.7 -> 83.3Improved
use evidence-aware reward instead of vanilla GRPO on OGVRF (ablation)Chain-of-Glimpse · Video-Holmesaccuracy (%): 41.8 -> 42.5Improved
add OGVRF for long videosQwen2.5-VL-7B · NExTQA Long videosaccuracy (%): 77.9 -> 83.2Improved
increase MCTS rollouts from 2 to 8 (ablation)Chain-of-Glimpse-7B · NExT-GQAAcc@GQA: 30.5 -> 32.1Improved
increase MCTS rollouts from 8 to 10 (ablation)Chain-of-Glimpse-7B · NExT-GQAAcc@GQA: 32.1 -> 30.9Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy