Welcome you! This is my implementation of a CNN Classifier in Pytorch for the Plant Pathology 2020 Competition hosted at Kaggle.
The goal of the Competition was to train a model using images of a training set to
- Accurately classify a given image from testing dataset into different diseased category of healthy leaf;
- Accurately distinguish between many diseases, sometimes more than one on a single leaf;
- Deal with rare classes and novel symptoms;
- Address depth perception—angle, light, shade, physiological age of the leaf; and
- Incorporate expert knowledge in identification, annotation, quantification, and guiding computer vision to search for relevant features during learning.
I was able to achieve the top 17% leaderboard with an MC AUROC Score of 0.97165, getting the 219° place (of a total of 1317 teams).
First, i performed a simple EDA on the dataset provided by Kaggle. You can check it in the notebook Exploratory Analysis.ipynb.
Second, in the notebook CNN Train in Pytorch.ipynb you shall find every bit of detail of how the training was accomplished using a ResneSt50 architecture. By the end of the entire training process, i was able to achieve a 95.968% accuracy on the test set.
I set a seed if you wish to reproduce the results. You can obtain the dataset in the original link of the competition that i provided at the beggining.