The README file will be an active document recording the updates. Eventually I will have a proper writeup.
The plan for this repository is to try out various experiments with the coffee disease detector model.
The goal is not to come up with a solution but to use various tools available for ML and DL development and use this project as a anchor project to try these tools.
First I will be using this project for ML experiment tracking using Dagshub. I want to try and use the MLFlow instance to track experiments and parameters as I run new experiments.
Some possible experiments that I will be trying is listed below:
-
Augment the data by increasing the number of positive and negative pairs for each class.
-
Use SimCLR suggestions for augmenting the dataset.
-
Try different contrastive functions.
More to come soon.