AI Paper Atlas

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

ICE: Inter-Instance Contrastive Encoding for Unsupervised Person Re-Identification

Hao Chen, Benoit Lagadec, François Bremond

ICCV 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Cross-modality and occlusion re-identification · 310 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 hard instance contrastive loss (ablation)camera-aware proxy baseline · Market1501mAP: 79.3 -> 80.5Improved
add soft instance consistency loss (ablation)camera-aware proxy baseline · Market1501mAP: 79.3 -> 81.1Improved
add both losses (ablation)camera-aware proxy baseline · MSMT17mAP: 36.4 -> 38.9Improved
add both losses (ablation)camera-aware proxy baseline · DukeMTMC-reIDmAP: 67.3 -> 69.9Improved
add hard instance contrastive loss (ablation)camera-agnostic baseline · Market1501mAP: 65.8 -> 78.2Improved
add soft instance consistency loss alone (ablation)camera-agnostic baseline · Market1501mAP: 65.8 -> 47.2Hurt
add soft instance consistency loss on top of hard loss (ablation)camera-agnostic ICE · Market1501mAP: 78.2 -> 79.5Improved
add soft instance consistency loss on top of hard loss (ablation)camera-agnostic ICE · MSMT17mAP: 30.3 -> 29.8Hurt
use all negatives instead of hardest negative in hard loss (ablation)ICE (aware) · Market1501mAP: 80.1 -> 82.3Improved
use KL with unaugmented target instead of MSE (ablation)ICE (aware) · Market1501mAP: 80.0 -> 82.3Improved
use KL with unaugmented target instead of strong-strong augmentation (ablation)ICE (aware) · Market1501mAP: 80.4 -> 82.3Improved
IBN-ResNet50 backbone instead of ResNet50ICE (aware) · MSMT17mAP: 38.9 -> 40.6Improved

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