This repository contains work done for the Meal Compliance Project, at the University of Canberra, for the Canberra Hospital.
The project is broken down into 6 sub projects
- Analysis - Random bits of code used to analyse model performance and prototype
- Additional Data Creator - applies image augmentation to create additional training data
- Image Preprocessor - This system is not complete, Given image of tray, crop and resize the image to the tray
- Image Segmentation - Working files for testing out the modified U-Net model
- SegmentMapMerger - Given the layer output from photoshop, and the classes database, create the annotated Y files for training and testing
- Integration Test - Flask API that serves the model. This contains the final files
See each project folder for more details
Note that this project uses openly licenced code from divamgupta