Improving Handwritten Text Recognition via 3D Attention and Multi-Scale Training

Zi-Rui Wang

The segmentation-free research efforts for addressing handwritten text recognition can be divided into three categories: connectionist temporal classification (CTC), hidden Markov…