Since 2010 Dr. Yeghiazaryan worked as a scientific researcher at Yerevan Physics Institute where he defended his PhD in Quantum Field Theory in 2012. Since 2016 he has been working in the field of Machine Learning, since 2018 Dr. Yeghiazaryan has been working in AI team of Webb Fontaine. Topics of his interest include Recommendation Systems, Computer Vision, NLP, others.

Reading text in natural images is gaining increasing attraction in the field of computer vision. Also the OCR of different types of documents has important business applications. Open source general purpose solutions like Tesseract can't be relied upon to work well for custom applications. Tailored solutions for these type of problems generally include two phases - text detection and text recognition. Both phases possess their unique challenges and require specialized techniques for solving them. In this talk I will discuss the general approaches and some new trends for solving these tasks.