Pitaya detection in orchards using the MobileNet-YOLO model
Xiuli Li, Yi Qin, Fujie Wang, Feng Guo, John T. W. Yeow
Abstract
The real-time detection and recognition of pitaya fruit is an important prerequisite for automatic picking. We combined with the current deep learning method with good recognition accuracy to realize the real-time detection and identification of pitaya fruit. Firstly, we collected a large number of pictures of pitaya fruit for labeling, and completed the production of data sets of Pitaya fruit. Then we use YOLOV3, YOLOV3-tiny and MobileNet-YOLO network models to train. After training, we test the performance of the trained model on the test data set. The experimental results show that the improved MobileNet-YOLO model has better detection speed than the YOLOV3 model, and the detection accuracy is better than the YOLOV3-tiny model. It can take into account the detection efficiency and accuracy, and detect the Pitaya fruit in the orchard in real time. Moreover, the MobileNet-YOLO model is a lightweight model, which can be deployed to the picking machine in the future, which can effectively provide Pitaya fruit detection and be applied to the actual environment of the orchard.