AI Paper Atlas

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

Selective Sparse Sampling for Fine-Grained Image Recognition

Yao Ding, Yanzhao Zhou, Yi Zhu, Qixiang Ye, Jianbin Jiao

ICCV 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 295 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 discriminative branch to original-only network (ablation)S3N O · CUB-200-2011total top-1 accuracy (%): 86.0 -> 87.6Improved
add complementary branch to original-only network (ablation)S3N O · CUB-200-2011total top-1 accuracy (%): 86.0 -> 87.3Improved
add complementary branch to O+D (ablation)S3N O+D · CUB-200-2011total top-1 accuracy (%): 87.6 -> 88.5Improved
add discriminative branch to O+C (ablation)S3N O+C · CUB-200-2011total top-1 accuracy (%): 87.3 -> 88.5Improved
add original branch to D+C (ablation)S3N D+C · CUB-200-2011total top-1 accuracy (%): 87.5 -> 88.5Improved
add discriminative branch (ablation)S3N O+C · CUB-200-2011top-1 localization accuracy (%): 56.6 -> 65.2Improved
add complementary branch (ablation)S3N O+D · CUB-200-2011top-1 localization accuracy (%): 59.2 -> 65.2Improved
use sparse attention instead of class-agnostic saliency attention (ablation)S3N · CUB-200-2011top-1 accuracy (%): 85.9 -> 88.5Improved
use sparse attention instead of dense class response attention (ablation)S3N · CUB-200-2011top-1 accuracy (%): 87.8 -> 88.5Improved
probabilistic partition instead of giving all attention to both branches (ablation)S3N · CUB-200-2011top-1 accuracy (%): 87.7 -> 88.5Improved
probabilistic partition instead of random partition (ablation)S3N · CUB-200-2011top-1 accuracy (%): 87.9 -> 88.5Improved
S3N versus ResNet-50 baseline (FT ResNet)ResNet-50 · CUB-200-2011top-1 accuracy (%): 86.0 -> 88.5Improved

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