AI Paper Atlas

Research map / Detection and segmentation / CNN features

Check and Link: Pairwise Lesion Correspondence Guides Mammogram Mass Detection

Ziwei Zhao, Dong Wang, Yihong Chen, Ziteng Wang, Liwei Wang

ECCV 2022 · Detection and segmentation › CNN features · approach: Efficient DETR query design

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 View-Interactive Lesion Detector (ablation)Deformable DETR · DDSMRecall@0.25 FPI (%): 73.8 -> 76.1Improved
add Lesion Linker alone, without VILD (ablation)Deformable DETR · DDSMRecall@0.25 FPI (%): 73.8 -> 74.1No clear effect
add Lesion Linker on top of VILD (ablation)VILD · DDSMRecall@0.25 FPI (%): 76.1 -> 78.1Improved
add Lesion Linker on top of VILD (ablation)VILD · DDSMRecall@2.0 FPI (%): 91.7 -> 92.4Improved
replace link-query Lesion Linker with pair verification (ablation)VILD · DDSMRecall@0.25 FPI (%): 76.1 -> 75.7No clear effect
replace link-query Lesion Linker with paired lesion query (ablation)VILD · DDSMRecall@0.25 FPI (%): 76.1 -> 68.8Hurt
use multiplicative instead of additive matching cost (alpha 0.5) (ablation)CL-Net · DDSMRecall@0.25 FPI (%): 73.8 -> 78.1Improved
use multiplicative instead of additive matching cost (alpha 0.5) (ablation)CL-Net · DDSMRecall@2.0 FPI (%): 90.0 -> 92.4Improved
vary alpha in multiplicative cost from 0.5 to 0.75 (ablation)CL-Net · DDSMRecall@0.25 FPI (%): 78.1 -> 74.8Hurt
vary alpha in multiplicative cost from 0.5 to 0.25 (ablation)CL-Net · DDSMRecall@1.0 FPI (%): 78.1 -> 76.1 at R@0.25; 88.0 -> 89.7 at R@1.0Mixed
increase Lesion Linker attention layers from 1 to 3 (ablation)CL-Net · DDSMRecall@0.25 FPI (%): 77.1 -> 78.1Improved
increase Lesion Linker attention layers from 3 to 5 (ablation)CL-Net · DDSMRecall@0.25 FPI (%): 78.1 -> 75.1Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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