Fig Fruit Recognition Method Based on YOLO v4 Deep Learning

Abstract

Image-based fig fruit recognition is a key technology to achieve smart fig planting management. However, compared with apples and mangoes, fig fruits are less different in color from the background and have more dense branches and leaves. This makes the detection of fig fruits more challenging. In this paper, we propose a fig recognition method based on YOLO v4 deep learning technique to achieve fast and accurate recognition and localization of fig fruits in complex environment images. This paper also compares the detection effect of YOLO v4 with Faster R-CNN and YOLO v3 algorithms, which were widely used in the field of fruit recognition in the past, on the same fig dataset. The experimental results show that the detection effect of the fig fruit recognition model constructed based on the YOLO v4 algorithm has improved to a certain extent in terms of average precision and other core metrics. It proves that YOLO v4 deep learning method has good detection effect on figs in complex environment and can provide technical support for intelligent fig planting management.