Abstract: Automation of cherry sorting not only aids in cherry industry's healthy development but also provides higher quality products to consumers．Thus, a cherry sorting system based on machine vision was designed．It achieved single cherry transporting through sorting cups, cherry images were captured by an industrial camera and combined with specially developed recognition software, OpenCV and NumPy libraries to extract external contours and measure sizes of cherries．Additionally, a corresponding UI interface was developed based on Thinker platform, embedding detection models such as YOLOv5 to achieve all-angle sorting detection of cherries．Results indicated that this method was simple and effective, accurately recognizing and measuring external dimensions of cherries．