How abhistacodes/handwritten-math-to-latex works
This repository is a small handwritten-math-to-LaTeX project. Its documented user flow is an equation image processed by a visual encoder, attention mechanism, and sequence decoder to produce LaTeX. The sampled runtime code also parses CROHME InkML and renders strokes, but no excerpt shows those helpers wired into the recognition flow. The README describes the recognition architecture; implementation details beyond the parser and renderer are not shown, so those nodes use the existing notebooks directory as their location.
Main entry points: notebooks Visual encoder
Components 8
Recognition
- Visual encoder
notebooks - Attention
notebooks - Sequence decoder
notebooks
InkML utilities
- InkML parser
src/data/inkml_parser.py - Stroke renderer
src/data/renderer.py
Other components
- Equation image
- LaTeX output
- CROHME InkML
Connections 4
- Equation image → Visual encoder: provides image
README.md - Visual encoder → Attention: encodes features
README.md - Attention → Sequence decoder: guides decoding
README.md - Sequence decoder → LaTeX output: generates sequence
README.md