A Comprehensive Survey of Methods and Applications for Detecting Infected Plant Leaves

Agriculture is the backbone of the Indian economy.Now that crop production is decreasing more quickly, farmers find it difficult to sell their goods when their output is decreased due to diseases that affect trees, particularly the leaves.To improve quality and productivity, it is imperative to treat any serious illnesses as soon as possible.This issue led to the creation of expert systems for disease prevention and innovative technologies for the detection and diagnosis of plant diseases.A detailed literature survey highlighted the applications of different architectures of deep learning models for identifying infected and healthy leaves.Most of the existing literature commonly emphasized on the use of machine learning techniques for activities like disease detection, fruit identification which are related to agriculture.This paper uses five deep learning models YOLOv8, Faster R-CNN, Swin Transformer, EfficientNet, and Custom CNN to classify whether the leaf is healthy or diseased.A dataset containing two classes of images of healthy and infected leaves has been used, and EfficientNet achieves the highest accuracy of 98.75% when compared to Faster R-CNN (87.98%),YOLOv8 (96.56%),Swin Transformer (97.00%) and Custom CNN (85.64%).However, more optimization is required, taking into account variables like dataset variations, computational performance, real-world applicability, and combining with IoT.