Document Reader using Deep Learning - Pedro Augusto Lopes Barbosa (Accenture)

Optical Character Recognition (OCR) has been around for a long time. This computer vision task aims to implement systems capable of reading textual content in a variety of different domains. Since the first implementations of OCR's systems in the 50', there were huge improvements in the areas of machine learning and computer vision with the most recent developments of deep learning techniques. We present a deep learning system for real time document reading using tensorflow research deep learning models. We show the problems and challenges we got through to implement our framework achieving great results with a small dataset.

Pedro is a machine learning/computer vision engineer at Accenture. I have BS in computer engineering and MSc in computer science focused in image processing.