Project for the course "Artificial Neural Networks and Deep Learning". Given a dataset containing multiple plant images, train a CNN able to classify correctly all the images. We implemented various advanced techniques like Augmentation, Outlier Detection, Transfer Learning, Fine Tuning, and we were able to get 93% accuracy, reaching top 5 performances between all the groups (over 170) partecipating the challenge.
Professor: Matteo Matteucci
Final Score: 5.5/5 (award for particularly good projects)