AI Paper Atlas

Research map / Video / Video segmentation and tracking

Advancing Complex Video Object Segmentation via Progressive Concept Construction

Zhixiong Zhang, Shuangrui Ding, Xiaoyi Dong, Songxin He, Jianfan Lin, Junsong Tang, Yuhang Zang, Yuhang Cao, Dahua Lin, Jiaqi Wang

ICLR 2026 · Video › Video segmentation and tracking · approach: Matching-based video object segmentation · 21 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 concept guidance on top of pixel-level association (ablation)SeC · SeCVOSJ&F: 62.2 -> 70.0Improved
add concept guidance on top of pixel-level association (ablation)SeC · SA-VJ&F: 82.4 -> 82.7Improved
add pixel-level association (long-term memory) to SAM 2.1 (ablation)SAM 2.1 · SA-VJ&F: 78.6 -> 82.4Improved
add pixel-level association (long-term memory) to SAM 2.1 (ablation)SAM 2.1 · SeCVOSJ&F: 58.2 -> 62.2Improved
construct concept offline from the full video instead of online (ablation)SeC · SeCVOSJ&F: 70.0 -> 71.8Improved
increase LVLM size from 1B to 4B (ablation)SeC · SeCVOSJ&F: 68.4 -> 70.0Improved
increase LVLM size from 4B to 8B (ablation)SeC · SeCVOSJ&F: 70.0 -> 70.3No clear effect
use 4 concept tokens instead of 1 (ablation)SeC · SeCVOSJ&F: 70.0 -> 69.9No clear effect
use HSV scene detector instead of absolute difference (ablation)SeC · SeCVOSJ&F: 69.0 -> 70.0Improved
use HSV scene detector instead of ORB (ablation)SeC · SeCVOSJ&F: 68.7 -> 70.0Improved
use HSV scene detector instead of optical flow (ablation)SeC · SeCVOSJ&F: 69.5 -> 70.0Improved
activate concept guidance on under 10% of frames rather than all frames (ablation)SeC · SeCVOSJ&F: 7.4% guidance, 70.0 J&FNo clear effect

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