AUTHOR=Jia Weikuan , Liu Mengyuan , Luo Rong , Wang Chongjing , Pan Ningning , Yang Xinbo , Ge Xinting 

TITLE=YOLOF-Snake: An Efficient Segmentation Model for Green Object Fruit

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2022.765523

DOI=10.3389/fpls.2022.765523

ISSN=1664-462X

ABSTRACT=Accurate detection and segmentation of object fruits is the basic premise of orchard yield measurement and automated picking, which britngs new challenges to efficient and accurate identification and segmentation of green object fruits in complex orchard environments due to the scenic range of factors such as light, weather, and operation angle. For the green fruit segmentation challenge, an efficient YOLOF-Snake segmentation model is proposed. First, the backbone network adopts the ResNet101 structure to realize the feature extraction of target fruits, and the obtained C5 feature map is expanded with the sense field, and classification and regression are performed with the help of decoder. The center point is taken in the regression frame to get the diamond-shaped frame. Finally, it is input to an additional deep-snake network and adjusted to the contour of the target fruit to achieve fast and accurate segmentation of green fruits. The experimental results show that the new model is sensitive to the green target fruit region, and the segmentation accuracy and efficiency are improved, which can effectively extend the application scope of agricultural equipment and provide theoretical reference for other fruit and vegetable target fruit segmentation.