AI Paper Atlas

Research map / Video / Video segmentation and tracking

General and Task-Oriented Video Segmentation

Mu Chen, Liulei Li, Wenguan Wang, Ruijie Quan, Yi Yang

ECCV 2024 · Video › Video segmentation and tracking · approach: Video semantic and panoptic segmentation · 19 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 shape-position aware query matching (ablation)GvSeg baseline · VIPSeg valVPQ: 36.0 -> 37.8Improved
add task-oriented query initialization and association (ablation)GvSeg · VIPSeg valVPQ: 37.8 -> 40.1Improved
add task-oriented temporal contrastive learning sampling (ablation)GvSeg · VIPSeg valVPQ: 40.1 -> 41.2Improved
combine all task-oriented components (ablation)GvSeg · VIPSeg valVPQ: 36.0 -> 44.0Improved
add shape and position cues to thing object association (ablation)GvSeg · VIPSeg valVPQ: 42.1 -> 44.0Improved
add shape and position cues to stuff object association (ablation)GvSeg · VIPSeg valVPQ: 42.1 -> 41.7Hurt
sample thing examples from whole video and stuff examples from dataset queue instead of nearby frames (ablation)GvSeg temporal contrastive learning · VIPSeg valVPQ: 40.1 -> 44.0Improved
increase queue length from 100 to 200 (ablation)GvSeg · VIPSeg valVPQ: 44.0 -> 44.1No clear effect
increase histogram bins beyond 36x12 (ablation)shape-position descriptor · VIPSeg valVPQ: 44.0 -> 43.9 (36x18), 44.0 (48x12)No clear effect
apply task-oriented design in general video segmentation on Occluded-VISGvSeg · Occluded-VIS val R-50AP: 31.1 -> 35.9 vs TarVISImproved

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