JONNY-ME/Microsoft-Rice-Disease-Classification-Challenge

Identifing disease types in images of rice grown in Egypt. # Microsoft-Rice-Disease-Classification-Challenge *Identifing disease types in images of rice grown in Egypt.* I used two different approaches for this challange. * Fastai * Pythorch Lightning **Fastai approach** ~ 0.042 9th place in private LB The file `fastai.ipynb` has this approach and It goes though all the processes starting from loading the data. I used different pretrained models from Vision transformer, Convnext and Swin-Transformer. all the models are found in here. For all the models a size of 224x224 and only the RGB images are used. **Lightning approach** ~0.056 in private LB The file `lightning.ipynb` has this approach. In this approach in addition to the fastai approach I tried to use the RGN images also but it didn't go well. here a single model with different folds prediction method is used. I haven't yet finshed the experment yet. Please Leave a⭐️ to my repo, if u find it useful😊