We present DeepPaperComposer, a simple solution for preparing highly accurate (100%) training data without manual labeling to extract content from scholarly articles using convolutional neural networks (CNNs). We used our approach to generate data and trained CNNs to extract eight categories of both textual (titles, abstracts, authors, headers, figure and table captions, and body texts) and nontextual content (figures and tables) from 30 years of 2916 IEEE VIS conference papers, of which a

more »

... were scanned bitmap PDFs. We curated this dataset and named it VISpaper-3K. We then showed our initial benchmark performance using VISpaper-3K over CS-150 using YOLOv3 and Faster-RCNN. We have opensourced DeepPaperComposer for training data generation 1 and have released the resulting annotation data VISpaper-3K 2 to promote reproducible research.