AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Streaming Convolutional Neural Networks for End-to-End Learning With Multi-Megapixel Images

Hans Pinckaers, Bram van Ginneken, Geert J. S. Litjens

TPAMI 2020 · Architectures and efficiency › CNN design · approach: Compact CNN architectures · 124 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 input resolution from 1024x1024 to 4096x4096 (ablation)Streaming CNN · TUPAC16 test setSpearman rho: 0.485 -> 0.536Improved
increase input from 2048x2048 to 4096x4096 with deeper network (ablation)Streaming CNN · TUPAC16 test setSpearman rho: 0.491 -> 0.570Improved
increase input from 4096x4096 to 8192x8192 (ablation)Streaming CNN · TUPAC16 test setSpearman rho: 0.570 -> 0.560No clear effect
increase input from 4096x4096 to 8192x8192 (ablation)Streaming CNN · TUPAC16 tuning setSpearman rho: 0.644 -> 0.692Improved
increase input resolution from 2048x2048 to 8192x8192 (ablation)Streaming CNN · CAMELYON17 macro-metastasisAUC: 0.515 -> 0.827Improved
increase input resolution from 2048x2048 to 8192x8192 (ablation)Streaming CNN · CAMELYON17 negativeAUC: 0.580 -> 0.706Improved
increase input resolution from 2048x2048 to 8192x8192 (ablation)Streaming CNN · CAMELYON17 isolated tumor cellsAUC: 0.450 -> 0.463No clear effect
increase input resolution from 2048x2048 to 8192x8192 (ablation)Streaming CNN · CAMELYON17 micro-metastasisAUC: 0.689 -> 0.709Improved
stream with 16 tiles instead of full imageconv layers · 1024x1024 inputpeak memory: 1957 MB -> 202 MBImproved
stream with 16 tiles instead of full imageconv layers · 1024x1024 inputbackward time: 45 ms -> 147 msHurt
train with streaming instead of conventional trainingsmall CNN · ImageNettetraining lossNo 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