AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Delving into Inter-Image Invariance for Unsupervised Visual Representations

Jiahao Xie, Xiaohang Zhan, Ziwei Liu, Yew-Soon Ong, Chen Change Loy

IJCV 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 62 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 inter-image invariance branch (InterCLR)BYOL, 200 epochs · ImageNet lineartop-1 accuracy: 70.6 -> 73.5Improved
add inter-image invariance branch (InterCLR)NPIDv2, 200 epochs · ImageNet lineartop-1 accuracy: 64.6 -> 65.7Improved
add inter-image invariance branch (InterCLR)MoCov2, 200 epochs · ImageNet lineartop-1 accuracy: 67.5 -> 68.0Improved
add inter-image invariance branch (InterCLR)BYOL, 1000 epochs · ImageNet lineartop-1 accuracy: 73.4 -> 74.5Improved
add inter-image invariance branch (InterCLR)MoCov2, 200 epochs · VOC07 low-shotmAP: 68.2 -> 70.7Improved
add inter-image invariance branch (InterCLR)BYOL, 200 epochs · ImageNet 1% labelstop-5 accuracy: 76.8 -> 79.4Improved
add inter-image invariance branch (InterCLR)BYOL, 200 epochs · VOC07+12 detectionAP50: 81.4 -> 82.7Improved
use online mini-batch k-means instead of offline global k-means for pseudo-labels (ablation)NPIDv2-InterCLR · ImageNet / VOC07 (empirical study)performanceImproved
use online labels instead of offline labels (training cost) (ablation)Intra+Inter · 8 V100 GPUs, 200 epochstraining time: 6.4 days -> 5 daysImproved
use semi-hard negative sampling instead of hard negative sampling (ablation)NPIDv2-InterCLR · VOC07 (empirical study)mAPImproved
use negative cosine margin m=-0.5 in inter-image branch instead of zero margin (ablation)NPIDv2-InterCLR · ImageNet linear (empirical study)top-1 accuracyImproved
use positive cosine margin in intra-image branch (ablation)NPIDv2 · ImageNet linear (empirical study)top-1 accuracyImproved

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