AI Paper Atlas

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

Grafit: Learning Fine-Grained Image Representations With Coarse Labels

Hugo Touvron, Alexandre Sablayrolles, Matthijs Douze, Matthieu Cord, Hervé Jégou

ICCV 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 84 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 instance loss to kNN loss (FC head) (ablation)Grafit FC · ImageNet-1kmAP (%): 30.7 -> 41.7Improved
add instance loss to kNN loss (FC head) (ablation)Grafit FC · CIFAR-100kNN top-1 accuracy (%): 70.0 -> 75.6Improved
add instance loss to cross-entropy (ablation)Baseline · ImageNet-1kkNN top-1 accuracy (%): 54.7 -> 65.0Improved
add coarse conditioning (ablation)Grafit FC · ImageNet-1kmAP (%): 41.7 -> 44.4Improved
add coarse conditioning (ablation)Grafit FC · CIFAR-100mAP (%): 53.6 -> 55.0Improved
add coarse conditioning (ablation)Baseline (cross-entropy) · ImageNet-1kmAP (%): 22.7 -> 24.4Improved
use MLP projector instead of FC (ablation)Grafit · CIFAR-100kNN top-1 accuracy (%): 75.6 -> 77.7Improved
use MLP projector instead of FC (ablation)Grafit · ImageNet-1kmAP (%): 44.4 -> 42.9Hurt
use oracle coarse labels at inference (ablation)Grafit · ImageNet-1kmAP (%): 42.9 -> 47.2Improved
add ClusterFit+ on top of baseline (ablation)ClusterFit+ · ImageNet-1kmAP (%): 22.7 -> 12.7Hurt
train Grafit with fine labels instead of coarse labelsGrafit · ImageNet-1kkNN top-1 accuracy (%): 69.1 -> 79.6Improved
replace baseline with Grafit for transferGrafit · iNaturalist 2019top-1 accuracy (%): 73.7 -> 75.9Improved

Later papers that compared with this method

2 later accepted papers report Grafit as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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