A Smartphone-Based Application for Scale Pest Detection Using Multiple-Object Detection Methods

Taiwan’s economy mainly relies on the export of agricultural products. If even suspicion a pest is found in crop products after they are exported, not only returned but whole batch crops destroyed, resulting extreme losses. The species mealybugs, Coccidae, and Diaspididae, which primary pests scale insect Taiwan, can lead to serious damage plants also severely affect production. Hence, recognize an important task field. In this study, we propose AI-based detection system for solving specific issue based pictures. Deep-learning-based object models, such as faster region-based convolutional networks (Faster R-CNNs), single-shot multibox detectors (SSDs), You Only Look Once v4 (YOLO v4), employed detect localize picture. experimental results show that YOLO achieved highest classification accuracy among algorithms, with 100% 89% 97% Diaspididae. Meanwhile, computational performance has indicated it suitable real-time application. Moreover, inference model further help end user. A mobile application using trained recognition been developed facilitate identification farms, helpful applying appropriate pesticides reduce