AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

ConCL: Concept Contrastive Learning for Dense Prediction Pre-training in Pathology Images

Jiawei Yang, Hanbo Chen, Yuan Liang, Junzhou Huang, Lei He, Jianhua Yao

ECCV 2022 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Data-efficient and robust self-supervision · 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
use dense (grid-level) contrasting, DenseCL instead of MoCo-v2MoCo-v2 · GlaS detectionAP bb: 52.3 -> 53.9Improved
use precise concept correspondence from reference view (g-ConCL s=5) instead of learned matchingDenseCL · CRAG detectionAP bb: 52.3 -> 55.5Improved
use bootstrapped K-Means concepts (b-ConCL) instead of image-level contrastMoCo-v2 · GlaS detectionAP bb: +4.5 AP bb and +3.1 AP mkImproved
use BASNet saliency concepts despite poor masksConCL · GlaS detectionAP bb: 52.3 -> 56.1Improved
add concept loss (lambda 0.0 -> 1.0) (ablation)b-ConCL · GlaS detectionAP bb: 52.3 -> 56.8Improved
add instance-loss warm-up (ablation)b-ConCL · GlaS detectionAP bb: 56.1 -> 56.8Improved
increase number of concepts K from 1 to 8 (ablation)b-ConCL · CRAG detectionAP bb: 50.0 -> 55.1Improved
increase K from 8 to 16 (ablation)b-ConCL · GlaS detectionAP bb: 56.8 -> 55.7Hurt
cluster from stage f5 instead of f4 (ablation)b-ConCL · CRAG detectionAP bb: 55.1 -> 56.5Improved
use ResNet-50 instead of ResNet-18 backbone (ablation)b-ConCL · GlaS detectionAP bb: 56.8 -> 57.0No clear effect
use b-ConCL pre-training with C4 detectorMask-RCNN-C4 · CRAG detectionAP bb: 49.4 -> 49.8Improved
use b-ConCL pre-training with RetinaNetRetinaNet · GlaS detectionAP bb: 47.2 -> 52.6Improved

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