Abstract: In order to improve the target recognition ability of unmanned plant protection machine and its adaptability to autonomous operation in complex environment, the machine learning algorithm is introduced into the design of the target autonomous recognition system of plant protection machine, and the accuracy of the recognition system is improved by using neural network learning algorithm and image enhancement processing technology. Simulate the working environment of the plant protection machine, set a large number of crop targets in the working area, and test its performance through the target recognition of the plant protection machine. The test results show that the plant protection machine can accurately identify the crop targets, so as to meet the design requirements of target independent recognition during independent operation.