Joint stroke classification and text line grouping in online handwritten documents with edge pooling attention networks

作者：

Highlights：

• A framework for joint text/non-text stroke classification and text line grouping in online handwritten documents is proposed.

• Stroke classification and text line grouping problems are formulated as node classification/clustering problems in graph.

• An edge pooling attention network (EPAT) is proposed to efficiently aggregate information of nodes and edges.

• Superior performance of both stroke classification and text line grouping is achieved on public datasets.

摘要

•A framework for joint text/non-text stroke classification and text line grouping in online handwritten documents is proposed.•Stroke classification and text line grouping problems are formulated as node classification/clustering problems in graph.•An edge pooling attention network (EPAT) is proposed to efficiently aggregate information of nodes and edges.•Superior performance of both stroke classification and text line grouping is achieved on public datasets.

论文关键词：Online handwritten documents,Stroke classification,Text line grouping,Graph neural networks,Edge pooling attention networks

论文评审过程：Received 10 July 2020, Revised 28 November 2020, Accepted 22 January 2021, Available online 2 February 2021, Version of Record 9 February 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.107859