Data and Code from: Smart vision-based monitoring system for AI-driven moth population estimation using camera-equipped trap imaging

Real-time, image-based monitoring for stored product insect pests could increase timely treatments and protection for postharvest products throughout the supply chain. Artificial intelligence (AI) and machine learning can provide the models necessary for accurate identification and population-counting within a trap-based system. This study presents the development of a smart vision-based monitoring system for moth population estimation using sticky traps with automated camera imaging. The proposed system integrates advanced image processing techniques with a Convolutional Neural Network (CNN) to accurately detect and classify moths and non-moth insects captured on sticky traps. Sticky traps, widely employed in integrated pest management (IPM) systems, often require manual inspection, which is labor-intensive and prone to human error. To address this, the developed system automates the detection process, reducing reliance on manual counting while improving classification precision. The dataset, consisting of 1,739 high-resolution images, was divided into training and testing subsets with a 70–30% split. Each image was preprocessed and annotated with ground-truth labels for accurate performance evaluation. The model demonstrated a high overall classification accuracy of 95.8%, with precision, recall, and F1-scores consistently exceeding 90%. These results highlight the effectiveness of the CNN in managing complex scenarios such as insect overlap, varied environmental conditions, and trap orientations, offering a scalable and efficient solution for real-time insect population monitoring in storage environments. The findings suggest that the proposed system provides a reliable and automated alternative for pest management, significantly reducing labor and enhancing decision-making in storage facilities and postharvest agriculture. In addition, field validation demonstrated the system’s feasibility in real-world storage environments, offering an effective and scalable alternative to traditional inspection practices while minimizing labor and enhancing precision pest control decisions.

This dataset features a subset of the images captured every hour in a sticky trap baited with a Plodia interpunctella pheromone lure. The images were processed for classificaiton of Indian meal moths and for population counting over time. A read me on image file naming convention, meta data, and conversion code for MatLab are included in the data files.

This research used resources provided by the SCINet project and/or the AI Center of Excellence of the USDA Agricultural Research Service, ARS project numbers 0201-88888-003-000D and 0201-88888-002-000D.