Comparison of Single-Shot and Two-Shot Deep Neural Network Models for Whitefly Detection in IoT Web Application

In this study, we have compared YOLOv4, a single-shot detector to Faster-RCNN, two-shot detect and classify whiteflies on yellow-sticky tape (YST). An IoT remote whitefly monitoring station was developed placed in rearing room. Images of attracted the trap were recorded 2× per day. A total 120 images labeled using labeling software split into training testing dataset, 18 additional yellow-stick with false positives increase model accuracy from monitors field that created due water beads reflective light after rain. The detection has two stages: region proposal then classification those regions refinement location prediction. Single-shot skips stage yields final localization content prediction at once. Because difference, YOLOv4 is faster but less accurate than Faster-RCNN. From results our it clear Faster-RCNN (precision—95.08%, F-1 Score—0.96, recall—98.69%) achieved higher level performance (precision—71.77%, score—0.83, recall—73.31%), will be adopted for further development station.