AI Paper Atlas

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

Unsupervised Learning of Dense Visual Representations

Pedro O O. Pinheiro, Amjad Almahairi, Ryan Benmalek, Florian Golemo, Aaron C. Courville

NeurIPS 2020 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Empirical contrastive visual learning · 50 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
Learn pixel-level contrastive representations rather than global image-level representationsVADeR vs. MoCo · PASCAL VOC12 val, frozen featuresmIoU: 43.0 -> 56.7Improved
Learn pixel-level contrastive representations rather than global image-level representationsVADeR vs. MoCo · NYU-depth v2 val, frozen featuresRMSE: 1.136 -> 0.964Improved
Learn pixel-level contrastive representations rather than global image-level representationsVADeR vs. MoCo · COCO val2017APbb: 38.9 -> 39.2Improved
Use correctly corresponding pixels instead of random unmatched pixel pairs (ablation)VADeR · PASCAL VOC12 val, frozen featuresmIoU: 32.4 -> 56.7Improved
Use correctly corresponding pixels instead of random unmatched pixel pairs (ablation)VADeR · DAVIS-2017 val, frozen featuresJ: 9.8 -> 52.4Improved
Use different crops rather than the same crop for paired views (ablation)VADeR · PASCAL VOC12 val, frozen featuresmIoU: 48.0 -> 56.7Improved
Use different crops rather than the same crop for paired views (ablation)VADeR · DAVIS-2017 val, frozen featuresJ: 46.7 -> 52.4Improved
Use VADeR initialization instead of supervised ImageNet initializationFine-tuning · PASCAL VOC12 val, 2% labelsmIoU: 37.4 -> 47.7Improved
Use VADeR initialization instead of MoCo initializationFine-tuning · PASCAL VOC12 val, 100% labelsmIoU: 75.3 -> 75.4No clear effect
Use VADeR initialization instead of ImageNet supervised initializationFine-tuning · NYU-depth v2 val, 2% labelsRMSE: 1.288 -> 1.128Improved

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