AI Paper Atlas

Research map / Detection and segmentation / Weakly supervised detection and segmentation

Parallel Detection-and-Segmentation Learning for Weakly Supervised Instance Segmentation

Yunhang Shen, Liujuan Cao, Zhiwei Chen, Baochang Zhang, Chi Su, Yongjian Wu, Feiyue Huang, Rongrong Ji

ICCV 2021 · Detection and segmentation › Weakly supervised detection and segmentation · approach: Weakly supervised semantic 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 instance-activation correlation loss L_ACL (ablation)PDSL · PASCAL VOC 2012 (ResNet18-WS)mAP: 17.8 -> 20.8Improved
add detection refinement heads (nr from 0 to 4) (ablation)PDSL · PASCAL VOC 2012 (ResNet18-WS)mAP: 8.7 -> 20.8Improved
increase nr from 4 to 5 (ablation)PDSL · PASCAL VOC 2012 (ResNet18-WS)mAP: 20.8 -> 20.9No clear effect
add class-specific mask heads (nm from 0 to 4) (ablation)PDSL · PASCAL VOC 2012 (ResNet18-WS)mAP: 16.0 -> 20.8Improved
increase nm from 4 to 5 (ablation)PDSL · PASCAL VOC 2012 (ResNet18-WS)mAP: 20.8 -> 20.8No clear effect
increase seed proposals nseed from 1 to 10 (ablation)PDSL · PASCAL VOC 2012 (ResNet18-WS)mAP: 15.1 -> 20.8Improved
increase nseed from 10 to 100 (ablation)PDSL · PASCAL VOC 2012 (ResNet18-WS)mAP: 20.8 -> 20.9No clear effect
learn detection and segmentation in parallel instead of cascade (ablation)PDSL · PASCAL VOC 2012mAPImproved
use PDSL with ResNet50-WS rather than ResNet18-WSPDSL · PASCAL VOC 2012mAP@0.50: 47.8 -> 49.6Improved
use PDSL instead of JTSM (top-down)JTSM · PASCAL VOC 2012mAP@0.50: 44.2 -> 47.8Improved
use PDSL instead of JTSM (top-down)JTSM · MS COCOmAP: 6.1 -> 6.3No 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