AI Paper Atlas

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

SCOUTER: Slot Attention-Based Classifier for Explainable Image Recognition

Liangzhi Li, Bowen Wang, Manisha Verma, Yuta Nakashima, Ryo Kawasaki, Hajime Nagahara

ICCV 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Self-explaining and visual explanation networks · 65 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
increase area-loss weight lambda from 1 to 10 (ablation)SCOUTER+ · ImageNet-100 positiveexplanation area size: 0.1561 -> 0.0476Improved
increase area-loss weight lambda from 1 to 10 (ablation)SCOUTER+ · ImageNet-100 positiveprecision: 0.8493 -> 0.9257Improved
increase lambda from 1 to 10 (ablation)SCOUTER+ · ImageNet-100 positiveDAUC: 0.1753 -> 0.2713Hurt
increase lambda from 1 to 10 (ablation)SCOUTER+ · ImageNet-100 classificationaccuracyNo clear effect
threshold heatmap (a>=0.2) to shrink explanation areaSS-CAM · ImageNet-100 positiveIAUC: 0.7143 -> 0.6010Hurt
threshold heatmap (a>=0.2) to shrink explanation areaSS-CAM · ImageNet-100 positiveDAUC: 0.4570 -> 0.7781Hurt
add GRU slot update instead of averaging (ablation)SCOUTER+ · ImageNet-100 positiveIAUC: 0.7456 -> 0.7647Improved
add GRU slot update instead of averaging (ablation)SCOUTER- · ImageNet-100 negativeIAUC: 0.5959 -> 0.6730Improved
add position embedding (ablation)SCOUTER- · ImageNet-100 negativeIAUC: 0.6141 -> 0.6730Improved
add position embedding (ablation)SCOUTER+ · ImageNet-100 positiveprecision: 0.8973 -> 0.9257Improved
replace FC classifier with SCOUTER+ResNeSt 26 · ImageNet-100accuracy: 0.8080 -> 0.7991Mixed
replace FC classifier with SCOUTER+ResNeSt 26 · CUB-200accuracy: 0.7538 -> 0.7362Hurt

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