MDCADNet: Multi dilated & context aggregated dense network for non-textual components classification in digital documents
作者：
Highlights：
• A novel multi-dilated densely connected network for chart images classification.
• A backend context module for effective intermediate features aggregation is proposed.
• The multi-resolution and larger receptive field modeling need for chart image scenarios is addressed.
• Extensive experiments using 7 benchmark datasets conducted for comparative analysis.
• Quantitative and qualitative results confirm the efficacy of the proposed model.
摘要
•A novel multi-dilated densely connected network for chart images classification.•A backend context module for effective intermediate features aggregation is proposed.•The multi-resolution and larger receptive field modeling need for chart image scenarios is addressed.•Extensive experiments using 7 benchmark datasets conducted for comparative analysis.•Quantitative and qualitative results confirm the efficacy of the proposed model.
论文关键词：Chart classification,Chart understanding,Multi dilation,Document intelligence,DenseNet
论文评审过程：Received 20 July 2021, Revised 10 November 2021, Accepted 18 January 2022, Available online 8 February 2022, Version of Record 16 February 2022.
论文官网地址：https://doi.org/10.1016/j.eswa.2022.116588
