Image Processing based ML Framework for Leaf Classification and Disease Detection

Image processing techniques are employed in this paper to supplement machine learning models for categorizing leaf species and detecting unhealthy spots in them. Multiple public datasets were mixed with a self-curated dataset of leaves from a homemade garden for the same. A set of 17 distinct features are extracted from background-subtracted leaf photos to capture information about texture, colour, and shape for leaf classification. These characteristics are used to train machine learning models such as Support Vector Machine and K-Nearest Neighbors, which do not require hardware intensive training. These models are trained on several dataset combinations, including the merged one, and then compared using precision, recall, and F1-score measures.