AI Paper Atlas

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

SERE: Exploring Feature Self-Relation for Self-Supervised Transformer

Zhong-Yu Li, Shanghua Gao, Ming‐Ming Cheng

TPAMI 2023 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Data-efficient and robust self-supervision · 19 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 SEREDINO ViT-S/16 · ImageNet-1K fine-tuningTop-1: 79.7 -> 80.9Improved
add SEREiBOT ViT-S/16 · ImageNet-1K fine-tuningTop-1: 80.9 -> 81.5Improved
add SEREiBOT ViT-B/16 · ImageNet-1K fine-tuningTop-1: 83.3 -> 83.7Improved
add SEREDINO ViT-S/16 · ImageNet-1K 1% labelsTop-1: 52.1 -> 55.9Improved
add SEREDINO ViT-S/16 · ImageNet-S pretrained initmIoU: 35.1 -> 36.9Improved
add SEREDINO ViT-S/16 · PASCAL VOC segmentationmIoU: 77.1 -> 79.7Improved
add SEREDINO ViT-S/16 · COCO detectionbox AP: 46.0 -> 46.6Improved
add SEREDINO ViT-S/16 · CIFAR-10 transferTop-1: 98.8 -> 98.9Improved
add SEREDINO ResNet-50 · PASCAL VOC (ImageNet-S300 pretrain)mIoU: 61.6 -> 62.5Improved
add spatial self-relation to image-level embedding (ablation)DINO · ImageNet-S300 pretrain, PASCAL VOCmIoU: 68.1 -> 70.7Improved
add channel self-relation to image-level embedding (ablation)DINO · ImageNet-S300 pretrain, PASCAL VOCmIoU: 68.1 -> 69.8Improved
add channel self-relation and L_I to spatial self-relation (ablation)SERE · ImageNet-S300 pretrain, PASCAL VOCmIoU: 71.5 -> 73.5Improved

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