Low resource handwritten text recognition

The project is taking handwritten word images as input and pass through a CNN followed by RNN. In the end, a CTC decoder is used to decode the text back into printed format. The detailed architecture is shown in the block diagram attached. The model is implemented, now the next task is to replace RNN with CNN for the sequence to sequence mapping of word characters.

Machine Learning (ML) Python Tensorflow Computer Vision Pytorch

프로젝트 ID: #30231273

프로젝트 소개