AI Paper Atlas

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

Spherical Feature Transform for Deep Metric Learning

Yuke Zhu, Yan Bai, Yichen Wei

ECCV 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 6 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 SFTNPair loss (GoogleNet) · Cars196Recall@1: 72.8 -> 79.4Improved
add SFTTriplet loss (GoogleNet) · CUB-200-2011Recall@1: 42.8 -> 48.3Improved
add SFTMulti-Similarity loss (GoogleNet-V2) · Cars196Recall@1: 84.0 -> 84.5Improved
add SFTRanked List Loss (GoogleNet-V2) · Cars196Recall@1: 74.2 -> 80.2Improved
add SFTTriplet, RLL, MS · SOPRecall@1: 70.8 -> 72.3; 77.5 -> 77.8; 73.1 -> 73.4No clear effect
use full SFT instead of degenerated SFT-d (ablation)NPair loss (GoogleNet) · Cars196Recall@1: 76.2 -> 79.4Improved
use SFT instead of FTL feature transformTriplet loss (GoogleNet) · Cars196Recall@1: 60.1 -> 65.1Improved
add SFTCosFace on MS-Celeb-1M subset · YTFverification accuracy (%): 96.2 -> 97.2Improved
add SFTCosFace · IJB-C TAR@FAR 0.001%TAR (%): 85.67 -> 87.19Improved
add SFT with unbalanced train scheme (ablation)Baseline face network · Long-tailed MS-Celeb-1M subset (tail classes)accuracy (%): 73.91 -> 82.32Improved
add SFT balanced train scheme (ablation)Baseline face network · Long-tailed MS-Celeb-1M subset (head classes)accuracy (%): 95.61 -> 95.77No clear effect
add SFT with accurate class center estimation (ablation)RLL on Cars196 · Cars196Recall@1: 74.2 -> 80.2Improved

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