AI Paper Atlas

Research map / Video / Video-language understanding

Think While Watching: Online Streaming Segment-Level Memory for Multi-Turn Video Reasoning in Multimodal Large Language Models

Lu Wang, Zhuoran Jin, Yupu Hao, Yubo Chen, Kang Liu, Yulong Ao, Jun Zhao

ECCV 2026 · Video › Video-language understanding · approach: Long video understanding · 14 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 segment memory notes (ablation)TWW multi-turn S3 · StreamingBench (Qwen3-VL-4B)accuracy: 52.35 -> 57.40Improved
use longer segments 120s/60s instead of 60s/30s (ablation)TWW multi-turn S3 · StreamingBenchaccuracy: 57.40 -> 55.33; tokens 302.56 -> 230.46Mixed
use shorter segments 30s/15s instead of 60s/30s (ablation)TWW multi-turn S3 · StreamingBenchaccuracy: 57.40 -> 57.20; tokens 302.56 -> 380.50No clear effect
add stage-3 long-range training over stage 2 (ablation)TWW single-turn · OVO-Bench (Qwen3-VL-4B)accuracy: 54.51 -> 55.02Improved
add stage-3 long-range training over stage 2 (ablation)TWW single-turn · StreamingBench (Qwen3-VL-4B)accuracy: 59.71 -> 60.04Improved
add streaming-aligned training to online Thinking (ablation)Qwen3-VL-4B Thinking · StreamingBenchaccuracy: 18.13 -> 60.04Improved
add streaming-aligned training (single-turn S3)Qwen3-VL-4B Thinking · OVO-Benchaccuracy: 50.70 -> 55.02Improved
use multi-turn segment memory instead of ThinkingQwen3-VL-4B · StreamingBenchavg output tokens: 689.22 -> 302.56Improved
use multi-turn segment memory instead of ThinkingQwen3-VL-4B · StreamingBenchaccuracy: 58.52 -> 57.40Hurt
use TWW pipeline instead of interleaved (ablation)Qwen3-VL-4B · StreamingBenchaccuracy: 55.35 -> 57.40Improved
use decoupled streaming pipeline instead of batch ThinkingQwen3-VL-4B · StreamingBenchTTFT: 31203.69 -> 2304.28Improved
add single-turn streaming S3 training for offline Video-MMEQwen3-VL-4B Thinking · Video-MMEaccuracy: 68.89 -> 73.41Improved

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