MIDV-2019: Challenges of the Modern Mobile-Based Document OCR

摘要

Recognition of identity documents using mobile devices has become a topic of a wide range of computer visionresearch. The portfolio of methods and algorithms for solving such tasks as face detection, document detection andrectification, text field recognition, and other, is growing, and the scarcity of datasets has become an important issue.One of the openly accessible datasets for evaluating such methods is MIDV-500, containing video clips of 50 identitydocument types in various conditions. However, the variability of capturing conditions in MIDV-500 did not addresssome of the key issues, mainly significant projective distortions and different lighting conditions. In this paper wepresent a MIDV-2019 dataset, containing video clips shot with modern high-resolution mobile cameras, with strongprojective distortions and with low lighting conditions. The description of the added data is presented, and experimentalbaselines for text field recognition in different conditions.