AI Paper Atlas

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

Scaling and Benchmarking Self-Supervised Visual Representation Learning

Priya Goyal, Dhruv Mahajan, Abhinav Gupta, Ishan Misra

ICCV 2019 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Invariance-based self-supervised learning · 363 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
scale pre-training data from 1M to 100M images (ablation)Jigsaw and Colorization (AlexNet, ResNet-50) · VOC07 classificationmAPImproved
use ResNet-50 instead of AlexNet as pre-training data grows (ablation)Jigsaw and Colorization · VOC07 classificationmAPImproved
increase Jigsaw permutations from 100 to 10k (ResNet-50) (ablation)Jigsaw, YFCC-1M · VOC07 classificationmAP: 5 point mAP improvementImproved
increase Jigsaw permutations from 100 to 10k (AlexNet) (ablation)Jigsaw, YFCC-1M · VOC07 classificationmAP: 1.9 point improvementImproved
vary Colorization soft-encoding neighbours K (ablation)Colorization, YFCC-1M · VOC07 classificationmAP: ~2 point mAP variationNo clear effect
pre-train Jigsaw on ImageNet-22k instead of ImageNet-1k (ablation)Jigsaw ResNet-50 · VOC07 detection (frozen conv body)mAP: 56.6 -> 67.1Improved
pre-train Jigsaw on ImageNet-22k instead of YFCC-100M (ablation)Jigsaw ResNet-50 · VOC07 detection (frozen conv body)mAP: 62.3 -> 67.1Improved
pre-train Jigsaw on YFCC-100M instead of ImageNet-22k (ablation)Jigsaw ResNet-50 · Places205 linear (layer4)top-1 accuracy: 41.5 -> 44.8Improved
pre-train Jigsaw on ImageNet-22k instead of YFCC-100M (ablation)Jigsaw ResNet-50 · VOC07 linear SVM (layer4)mAP: 71.0 -> 71.9No clear effect
replace random-init (scratch) with Jigsaw YFCC-100M pre-trainingResNet-50 PSPNet · NYUv2 surface normalsmean angle error: 26.3 -> 22.4Improved
replace Places205 supervised with Jigsaw YFCC-100MResNet-50 PSPNet · NYUv2 surface normalspixels within 11.25 degrees: 41.8 -> 44.6Improved
replace ImageNet-1k supervised with Jigsaw ImageNet-22kResNet-50 · VOC07 linear SVM (layer4)mAP: 80.4 -> 71.9Hurt

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