AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Efficient Classification of Very Large Images With Tiny Objects

Fanjie Kong, Ricardo Henao

CVPR 2022 · Architectures and efficiency › CNN design · approach: Compact CNN architectures · 33 citations (OpenAlex)

Check 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
Replace one-stage ATS sampling with two-stage hierarchical attention samplingZoom-In Network · Colon Canceraccuracy: 90.7 ± 1.4 -> 95.0 ± 2.6Improved
Replace one-stage ATS sampling with two-stage hierarchical attention samplingZoom-In Network · NeedleCamelyonaccuracy: 72.5 -> 76.0Improved
Use ResNet16 instead of LeNet as the feature-extractor backboneZoom-In Network · NeedleCamelyonaccuracy: 76.0 -> 78.1Improved
Use ResNet16 instead of LeNet as the feature-extractor backboneZoom-In Network · Traffic Signs Recognitionaccuracy: 91.2 -> 92.6Improved
Use ResNet16 instead of LeNet as the feature-extractor backboneZoom-In Network · Functional Map of the Worldaccuracy: 72.9 -> 74.3Improved
Use ResNet16 instead of LeNet as the feature-extractor backboneZoom-In Network · Camelyon16accuracy: 81.3 -> 82.6Improved
Add pixel-level annotations to trainingZoom-In Network · Camelyon16accuracy: 81.3 -> 88.2Improved
Add contrastive learning (ablation)Zoom-In Network · Colon Canceraccuracy: 94.0 -> 95.0Improved
Add contrastive learning (ablation)Zoom-In Network · NeedleCamelyonaccuracy: 74.3 -> 76.0Improved
Add contrastive learning (ablation)Zoom-In Network · Camelyon16accuracy: 80.6 -> 81.3Improved
Set entropy regularization λ to 0 instead of 1e-5 (ablation)Zoom-In Network · Colon Canceraccuracy: 95.0 -> 94.0Hurt
Set entropy regularization λ to 1.0 instead of 1e-5 (ablation)Zoom-In Network · Colon Canceraccuracy: 95.0 -> 95.0No clear effect

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