Handwritten English word recognition using a deep learning based object detection architecture

Handwriting is used to distribute information among people. To access this for further analysis the page needs be optically scanned and converted machine recognizable form. Due unconstrained writing styles along with connected overlapping characters, handwriting recognition remains a challenging task. Most of methods in literature use lexicon-based approaches train their models on large datasets having near 50 K word samples achieve good results. This results high computational requirements. While these around words dictionary when recognizing handwritten English text, actual number much higher than this. end, we propose technique recognize text based YOLOv3 object model that lexicon-free performs sequential character detection identification low training (only 1200 images). works well without any dependency writers’ style writing. tested IAM dataset it able 29.21% Word Error Rate 9.53% Character predefined vocabulary, which par state-of-the-art models.