Harnessing AI for Agriculture: Plant Pest Detection on Web Application Using Deep Learning

This research develops a mobile web application for pest identification using deep learning to help farmers manage pest outbreaks. The system enables farmers to capture pest images with their mobile cameras and receive alerts and recommendations. The objectives include developing pest classification models, creating web applications, and assessing user satisfaction. YOLOv5 architecture was tested for accuracy, speed, and efficiency. Developed using Flask and Python with SQLite for data management, results showed YOLOv5m6 achieved the highest Recall value of 0.8281 in 4.007 hours, followed by YOLOv5m at 0.8234 in 3.622 hours, and YOLOv5l at 0.8261 in 19.424 hours. Despite similar Recall values, YOLOv5m6 was more efficient. The system demonstrates AI's potential to improve pest management, though challenges in real-time processing and image complexity remain.