Description of Invention
Computational botany is a field of study where technologies are increasingly researched and developed for the benefit of plant related industries. However, current state-of-the-art automated plant identification require large dataset to ensure a reliably trained system. This work sets out to investigate the potential of synthetically generated fine-grained leaf images as the training data of classification models to classify real images in order to handle the data insufficiency issue. This project investigated the impact of synthesized image quality and quantity along with GAN model training length at producing reliable classifiers from limited real image data.