Abstract: In order to solve the problem that citizens, lack professional knowledge of garbage sorting , can not accurately judge the type of garbage, which leads to the unsatisfactory effect of garbage sorting, this paper introduces a system that can independently identify the type of garbage. This system uses visual recognition technology to compare the collected images with the trained model, calculates the garbage types with the highest similarity through the algorithm, and sends the results to the control center. The control center works according to the garbage type control delivery module. After testing, the range of recognition includes more than ten kinds of garbage common in all four categories and their different forms, and the recognition accuracy is about 78%.

Key words: intelligent garbage classification, garbage classification data set, visual recognition, delivery system