AI Paper Atlas

Research map / Detection and segmentation / CNN features

Character Region Attention For Text Spotting

Youngmin Baek, Seung Shin, Jeonghun Baek, Sungrae Park, Junyeop Lee, Daehyun Nam, Hwalsuk Lee

ECCV 2020 · Detection and segmentation › CNN features · approach: Arbitrary-shape scene text detection · 74 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 Character Region Attention (ablation)CRAFTS · IC13 end-to-endH-mean: 91.9 -> 92.2 (+0.3)Improved
Add Character Region Attention (ablation)CRAFTS · IC15 end-to-endH-mean: 73.2 -> 74.9 (+1.7)Improved
Add Character Region Attention (ablation)CRAFTS · TotalText end-to-endH-mean: 77.5 -> 78.7 (+1.2)Improved
Propagate recognition loss to detection stage (ablation)CRAFTS · ICDAR2015character error length: 1,251 -> 1,147 (-8.3%)Improved
Propagate recognition loss to detection stage (ablation)CRAFTS · TotalTextcharacter error length: 5,050 -> 4,521 (-10.4%)Improved
Add orientation estimation (ablation)CRAFTS · IC15 end-to-endH-mean: 74.1 -> 74.9Improved
Add orientation estimation (ablation)CRAFTS · TotalText end-to-endH-mean: 77.5 -> 78.7Improved
Replace VGG-16 CRAFT detector with ResNet50-based CRAFTSCRAFT detector · inference speedFPS: 8.6 FPS vs 8.3 (1280 side)Improved
Reduce input longer side from 2560 to 960CRAFTS · inference speedFPS: 5.4 -> 9.9Improved
Pretrain recognizer with other multi-language datasets with detector frozenCRAFTS · IC19-MLTend-to-end H-meanImproved
Use CRAFTS vs MaskTextSpotter on generic E2ECRAFTS · IC13 end-to-end genericH-mean: 88.2 -> 92.2Improved
Use CRAFTS vs CharNet on IC15 detectionCRAFTS · IC15 detectionH-mean: 91.5 -> 87.1Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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