[Python] How to study user handwriting styles on canvas (web) and also generate handwriting...

I am developing a web-based canvas application using Excalidraw for drawing and annotation, with a Python backend. My application currently detects drawn objects representing words, but I need to further analyze user handwriting and generate replacement strokes that match the user's style when correcting recognized words.

I have two main technical challenges:

Handwriting Feature Extraction:

I want to identify and extract measurable features from user-drawn strokes (e.g., stroke direction, curvature, and pressure) on the HTML5 canvas.

Question: What libraries or algorithms exist (for example, in JavaScript or Python) that can help extract these handwriting features from canvas input? Are there established techniques or research papers that detail this process?

Handwriting Stroke Generation:

I want to render corrected text onto the canvas by generating new strokes that mimic the user’s original handwriting style.

Question: What approaches—either machine learning, deep learning, or procedural methods—can convert text into stroke sequences that resemble a given handwriting style? Are there libraries or models available that demonstrate this capability? Any references to research, libraries, or example implementations that address either or both challenges would be greatly appreciated.

Continue reading...