Kaggle research code competition chosen as the final capstone project for the "Applied Machine Learning Course" at Stevens during my Masters. Read the abstract below:
Apple orchards in the United States are under constant threat from many pathogens and insects. Appropriate and timely deployment of disease management depends on early disease detection. Incorrect and delayed diagnosis can result in either excessive or inadequate use of chemicals, with increased production costs and increased environmental and health impacts.
We used the expert-annotated data set of 18,600 apple leaf images provided to the Kaggle community for the Plant Pathology Challenge as part of the Fine-Grained Visual Categorization(FGVC) workshop. Foliar disease labels included apple scab, cedar apple rust, complex diseases, frog eye leaf spot, powdery mildew, and healthy leaves. Participants were asked to use the image data set to train a machine learning model to classify disease categories and develop an algorithm for disease severity quantification. We trained a baseline sequential CNN model and achieved a validation set accuracy and F-1 Score of 83% and 77%respectively. This was our best model as opposed to two other classic CNN architectures (Transfer Learning) fine-tuned for this training set. InceptionResNetV2 and EfficientNetB1 achieved a validation F1-score of 68.74% and 73.02% respectively.
This data set will contribute toward development and deployment of machine learning–based automated plant disease classification algorithms to ultimately realize fast and accurate disease detection. The organization will continue to add images to the pilot data set for a larger, more comprehensive expert-annotated data set for future Kaggle competitions and to explore more advanced methods for disease classification and quantification.