Abstract: Aiming at the requirement of intelligent recognition of image targets by the storage area inspection image acquisition equipment, a vision-based long-distance suspicious target recognition algorithm was designed and implemented. Firstly, a target detection method was used to identify and collect the target image. And then, the convolution layer of the deep learning model based on convolutional neural network was used to extract the features of the target image, and the shallow network based on the traditional machine learning method was used to classify the suspicious target. Finally, an experiment was designed according to the algorithm. The experimental results show that the algorithm model can improve recognition effect, can effectively reduce the workload of manual recognition, and can meet the requirements of actual application.