DEFUSE: Deep Fused End-to-End Video Text Detection and Recognition

Detecting and recognizing text in natural scene videos images has brought more attention to computer vision researchers due applications like robotic navigation traffic sign detection. In addition, Optical Character Recognition (OCR) technology is applied detect recognize on the license plate. It will be used various commercial such as finding stolen cars, calculating parking fees, invoicing tolls, or controlling access safety zones aids detecting fraud secure data transactions banking industry. Much effort required when are low contrast motion blur with arbitrary orientations. Presently, detection recognition approaches limited static horizontal approximately text. dynamicity challenging because of presence multiple blurs caused by defocusing, motion, illumination changes, arbitrarily shaped, occlusion. Thus, we proposed a combined DeepEAST (Deep Efficient Accurate Scene Text Detector) Keras OCR model overcome these challenges proffered DEFUSE Fused) work. This two-combined technique detects regions then deciphers result into machine-readable format. The method experimented three different video datasets ICDAR 2015, Road 1K, own Datasets. Our results proved effective precision, recall, F1-Score.