This paper is under review which means review has begun. You can track the progress of this review on GitHub over here »

The repository provides software to train a neural network to detect text in screen images. The repository also provides code that generates training data by extracting figures and text masks from research papers.

Archive DOI: pending