InsectInsight - Using AI to detect insect on farm

This project develops an advanced pest detection system using deep learning to identify five harmful insects in agriculture. The team will process 20,000 images to create a robust dataset, then train and compare YOLO v8, v9, and v10 models. YOLO v9 was selected as the optimal model, balancing inference speed, detail capture, and performance metrics. Despite a slightly lower mAP50 than v8, v9's more complex architecture (about 3x more layers) suggests greater potential for task-specific optimization.

To overcome limitations with a developing luring device, the project simulates its camera view by superimposing insect images onto sample backgrounds. A full-stack mobile application will be created for farmers and administrators, featuring real-time insect population visualization, luring device monitoring, and on-device insect detection using phone cameras.

The app will support both Android and iOS platforms, with backend deployment on AWS for scalability. This system aims to revolutionize pest management in agriculture by providing farmers with an efficient, early detection tool. By combining advanced AI techniques with practical, user-friendly features, the project seeks to significantly improve crop protection strategies and overall agricultural productivity. The comprehensive approach integrates cutting-edge technology with real-world agricultural needs, potentially transforming how farmers monitor and manage pest infestations in their crops