Smart Farming and Crop Protection by Evaluating the Performance of Convolutional Neural Networks and YOLOv4 for Plant Leaf Disease Detection

Agriculture plays a significant role in India due to population growth and increased food demands. Hence, there is a need to enhance the yield of crops. Vegetation is frequently susceptible to a wide range of diseases that arise due to various seasonal and environmental conditions. These plant diseases not only jeopardize the quality and quantity of agricultural produce but also pose serious threats to farmers’ livelihoods and overall food security. Traditionally, the identity and treatment of plant diseases have relied closely on guide inspection and professional understanding, which may be time-consuming and susceptible to human mistakes. With recent advancements in technology, there is a growing interest in automated disease detection systems that leverage artificial intelligence and machine learning techniques. These contemporary solutions provide faster, more accurate and cost-effective techniques for identifying plant diseases, permitting farmers to take properly timed preventive and corrective measures. This study presents a novel approach to plant leaf disease detection and severity classification by leveraging the capabilities of YOLOv4 and Convolutional Neural Networks (CNNs). These machine learning algorithms have proven great potential in image processing and pattern recognition tasks, making them appropriate for diagnosing plant situations from visual information. We have used a dataset containing images of four various plant species, each suffering from different kinds of infections. By training these models by available datasets, the proposed system can recognize and classify diverse plant diseases with high accuracy. The performance parameters are evaluated extensively and results are derived. The accuracy of the CNN and YOLOv4 obtained around 95.5% & 91.0%, respectively.