Graph Neural Network based Handwritten Trajectories Recognition

One of the Artificial Intelligence (AI) important applications is human handwritten text recognition. The Handwriting Recognition (HWR) refers to recognizing handwriting through machines. The handwritten text scanned and recognized is offline HWR in nature, whereas recognizing while writing is online HWR Singh et al. [2017]. The handwriting trajectory refers to handwritten strokes which are set of sequential pixels in online HWR and set of pixels in offline HWR Pan et al. [2019]. In offline HWR, these trajectories writing orders recovered through drawing order techniques, and in online HWR, the digital pen strokes refer to trajectories Sharma [2015]. In either case, these trajectories are important sources of information to understand and recognize handwriting. Handwriting trajectories are also understood as the paths traced by a writing pen or stylus movements across a writing surface.