In the era of remote sensing big data, synthetic aperture radar (SAR) image interpretation is a great scientific application challenge [1]. The automatic detection and identification of targets are the hot spots and focuses of this research field. In general, the SAR image detection and recognition method require multiple independent steps such as filtering, edge extraction, image segmentation and target recognition. The complexity of the process not only reduces the efficiency of target detection and recognition of SAR images, but also loses a lot of key image features to a certain extent, which makes it difficult to further process SAR images and restrain the accuracy of detection and recognition. Therefore, this paper presents a method for automatic detection and recognition of SAR image targets based on convolution neural network (CNN). By using the CNN model to conduct target detection and recognition experiment on the moving and stationary target acquisition and recognition (MSTAR) benchmark dataset, we verify the validity of the deep neural network in the field of SAR image target recognition, which lays the foundation for further research in this field.