Identifying rare desert plants remains challenging due to the unavailability of large datasets of rare species. As of today, classical deep-learning models require a lot of training data. To address this issue, this project aims to develop a plant classifier using a Neural Network model using the Few-Shot Learning (FSL) method. FSL is particularly known for its effectiveness on small datasets, hence we employed it here to train the neural network on a small dataset of rare plants, resulting in high classification accuracy for these elusive plants.
The trained model will be implemented in our custom Android application, enabling users to capture or select a photo for classification. The application utilizes a cloud database to store various user data, including login credentials, classification history, and more.
