Insect Pest Management in Artificial Intelligence: Enhancing Insect Detection and Classification

Recognition, classification, and management of insect pests are crucial to preventing substantial crop losses.Traditional manual techniques are time-consuming, labor-intensive, and susceptible to errors, particularly in intricate agricultural environments.Traditional approaches frequently fail to accommodate dynamic pest behaviors, leading to crop losses and heightened chemical use.Consequently, the implementation of Artificial Intelligence (AI) methodologies in pest detection and management serves as an effective alternative to the issues presented by changing insect populations and the pursuit of sustainable agricultural methods.Therefore, this paper proposes an AI-driven method for lightweight and high-accuracy identification of insects and pests, designed for real-world deployment.The system employs deep learning (YOLOv11) for real-time insect and pest identification, utilizes pre-trained models (ResNet50 and VGG19) for feature extraction, and applies machine learning (Logistic Regression) for final classification to balance performance with computational efficiency.Additionally, we designed an offline insect identification software interface that can be installed on mobile devices for automated insect and pest identification, enhancing field usability.The system was trained and assessed using a composite dataset of 6,548 images from IP102 and insect classification datasets, categorized into 31 classes.Experimental results demonstrated that the proposed system can identify insects and pests with an accuracy of up to 95.19%.This lightweight AI system provides a pragmatic, scalable solution for agricultural surveillance.It offers quick and precise pest management suggestions, enhancing agrarian crops, minimizing pesticide misuse, and alleviating environmental harm.