Machine Learning Elements

Motivation

The goal is to provide users with the ability to upload hand-drawn graphs to the application, which can then be used within the application.

Tech stack / development environment

• Model training: Python PyTorch with Detectron2 in Google Colab
• Model execution: tensorflowjs in the browser of the user

Known problems / Open tasks

• Conversion of the model from pytorch/detectron2 (to tensorflow) to tensorflowjs
• Adjust post-processing after model recognition to the graph model used by the application
• Adjust pre-processing of the image according to the needs of the converted model (resizing? rgb(a)?)
• Find (cloud) storage to save the converted model for downloading to the browser of the user (git might not be the best place for large binary files)