AI Paper Atlas

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

Unifying Deep Local and Global Features for Image Search

Bingyi Cao, André Araujo, Jack Sim

ECCV 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 412 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 stop-gradient from attention and reconstruction losses to the backbone (ablation)DELG joint model (R50) · ROxf HardmAP: 42.4 -> 45.1Improved
add stop-gradient from attention and reconstruction losses to the backbone (ablation)DELG joint model (R50) · RPar MediummAP: 78.9 -> 81.6Improved
replace softmax loss with ArcFace loss (ablation)SPoC global feature · ROxf MediummAP: 51.2 -> 59.8Improved
replace SPoC pooling with GeM pooling (ablation)ArcFace global feature · ROxf MediummAP: 59.8 -> 69.3Improved
train global and local heads jointly without gradient control (ablation)GeM+ArcFace global feature · RPar MediummAP: 82.2 -> 78.9Hurt
add the local-feature head with stop-gradient (jointly trained) vs separate global model (ablation)GeM+ArcFace global feature · ROxf MediummAP: 69.3 -> 69.7No clear effect
add stop-gradient (ablation)naive joint training · activation sparsity of conv4 and conv5sparsity: 80% and 97% -> 45% and 88%Improved
increase autoencoder reconstruction weight lambda from 0 to 10, then 20 (ablation)autoencoder dimensionality reduction · GLD image-pair matchingAPMixed
add local feature re-ranking with spatial verificationR101-DELG · ROxf+1M HardmAP: 30.3 -> 39.3Improved
add local feature re-rankingR101-DELG · GLDv2-recognitionmicroAP@1: 32.0 -> 61.2Improved
binarize local features (DELG star)R101-DELG · ROxf+1M HardmAP: 39.3 -> 38.5Hurt
use 7 local scales instead of 3R101-DELG re-ranking · ROxf MediummAP: 77.5 -> 78.5Improved

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