Analyzing and Comparing Deep Learning Models for Strawberry Leaf Disease Identification and Classification

Abstract

Strawberry is highly prone to numerous foliar diseases which can drastically decrease the yield and quality unless well identified. The conventional ways of disease identification are based on manual inspection, which is very tedious and ineffective in large farms. The creation of artificial intelligence (AI) has enabled the application of deep learning (DL) models to identify illnesses in plants through the examination of images automatically. This paper will give a detailed comparison and analysis of various deep learning models used to diagnose and classify strawberry leaf diseases. A collection of images of normal and diseased strawberry leaves is used with a number of more advanced convolutional neural network (CNN) models, such as VGG16, ResNet50, and InceptionV3, and transfer learning is used to use the knowledge gained with pre-trained models and simplify the training process. The simulated results indicate that each of the models is a reasonable classification model, with InceptionV3 being the most accurate and VGG16 being the best tradeoff between accuracy and resource-efficiency, so it is applicable in real-time and resource-constrained conditions.