An Automatic Insect Identification and Pesticides Recommendation in Tea Plantation Using Machine Learning

Insect infestations in tea plantations result in significant crop losses, impacting both quality and yield. This paper proposes an automated system that utilizes a custom-trained YOLOv5 model to detect harmful insect species and recommend suitable pesticides. The model is trained on an annotated dataset prepared using Roboflow, ensuring high accuracy across various environmental conditions. Upon insect detection, the system references a CSV file containing pest-to-pesticide mappings to provide accurate recommendations. This approach reduces manual effort, minimizes pesticide misuse, and enhances pest management efficiency in tea plantations.