A Unified Survey of Supervised, Unsupervised, and Semi-Supervised Learning Techniques for Plant Leaf Disease Detection

Abstract The detection and classification of plant leaf diseases is critical for ensuring sustainable agricultural productivity. This survey presents a unified and comprehensive overview of supervised, unsupervised, and semi-supervised learning techniques applied to plant leaf disease detection and classification. Supervised approaches such as Convolutional Neural Networks (CNNs), Support Vector Machines (SVMs), and ensemble models have demonstrated high accuracy but require extensive labeled datasets. Unsupervised methods, including K-means clustering, autoencoders, and anomaly detection algorithms, offer promising results with unlabeled data, especially in early-stage disease detection. Meanwhile, semi-supervised learning bridges the gap by leveraging limited labeled and abundant unlabeled data through frameworks such as self-training, GANs, and hybrid models. This paper compares model accuracies, datasets used, tools and platforms (Python, TensorFlow, MATLAB), and evaluation metrics across recent literature. Emphasis is placed on the role of machine learning in real-time disease monitoring, data augmentation, and resource-efficient farming. The study provides practical insights and future directions for researchers and agritech developers aiming to integrate AI-driven solutions into precision agriculture. Keywords: Plant Leaf Disease Detection , Supervised Learning , Unsupervised Learning, Semi-Supervised Learning, Precision Agriculture