AI Paper Atlas

Research map / Detection and segmentation / CNN features

PageNet: Towards End-to-End Weakly Supervised Page-Level Handwritten Chinese Text Recognition

Dezhi Peng, Lianwen Jin, Yuliang Liu, Canjie Luo, Songxuan Lai

IJCV 2022 · Detection and segmentation › CNN features · approach: Arbitrary-shape scene text detection · 32 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 semantic matching (AR and edit distance) over page-level length check (ablation)PageNet · ICDAR13AR*: 15.18 -> 92.83Improved
add semantic matching over line-level length check (ablation)PageNet · ICDAR13AR*: 41.90 -> 92.83Improved
add spatial matching (ablation)PageNet · ICDAR13AR*: 91.62 -> 92.83Improved
use proposed pseudo-label updating instead of Replace (ablation)PageNet · ICDAR13AR*: 12.24 -> 92.83Improved
use proposed updating instead of Average (ablation)PageNet · ICDAR13AR*: 91.58 -> 92.83Improved
use proposed updating instead of Fixed Ratio (ablation)PageNet · ICDAR13AR*: 91.51 -> 92.83Improved
add pretraining stage on top of training (ablation)PageNet · ICDAR13AR*: 80.17 -> 91.57Improved
add initializing stage after pretraining (ablation)PageNet · ICDAR13AR*: 91.57 -> 92.83Improved
weak supervision instead of full supervision with character boxes (ablation)PageNet · ICDAR13AR*: 91.37 -> 92.83Improved
weak supervision instead of full supervision (ablation)PageNet · MTHv2AR*: 93.81 -> 93.76No clear effect
train on rotated pages with weak supervision vs full (ablation)PageNet · ICDAR13 multi-directionalAR*: 89.43 -> 89.52No clear effect

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