Abstract: A grapefruit classification detection system based on BP neural network was introduced. BP neural network model was built by selecting the grapefruit surface defects, grapefruit shape, grapefruit size and other indicators, in order to get better network weights. Then 20 sets of data for verification were selected. The experimental results indicated that grapefruit classification could be accurately classified by using BP neural network with the accuracy of 90%.

Key words: BP neural network, grapefruit classification, grapefruit characteristics