Project repository with the objective to train a model to be able to classify a given image of a leaf into one of the following classes: "Healthy", "Rust", "Scab" or "Multiple Diseases".
- This project is a submission for the plant-pathology-2020-fgvc7 challenge on the Kaggle website.
- The approach is an ensemble of multiple network architectures (MobileNet, EfficientNet, InceptionNet, ResNet and DenseNet).
- This project was carried out as part of the "Deep Learning" course on the RWU, supervised by Mark Schutera (@schutera).
- The result is a score of 96.6% on Kaggle's public leaderboard which is in the top 38% of all competitors.
- Details and results can be found here.