Plant Disease Detection using Pre-Trained Deep Learning Models: A Study on Low-Quality Images
Abstract
Plant disease is a big issue in development of agriculture. Protecting food security is addressing the challenge of feeding 5 billion people by 2050. Plant disease can impact food production and quality. Thus, early detection of plant disease automatic detection based on deep learning would save the quality and production of foods also improve the quality of foods and minimize the losses of yield. Researcher works with previous models have been applied for plant to high accuracy to require high quality images for classification and clustering but due to low quality images not effective. By using deep learning model and pre trained model based on convolution neural network to address these problems and early plant disease detection. Dataset includes both diseased and healthy crops leaves including 3 classes. Our goal is to ensure the ability of training dataset to generalize the performance of real word scenarios. This work represents an important step of advancing plant disease detection and optimizing dataset in deep learning application.