Abstract: Combining deep learning into target detection algorithm breaks through the performance bottleneck of traditional algorithms and becomes a hot research topic in computer vision field. the most popular object detection algorithm based on deep learning is studied deeply, and an overall understanding is obtained, which provides useful theoretical guidance for the advancement and efficiency of object detection application system development. The development process of deep convolutional neural network into object detection algorithm is sorted out in chronological order, and the main algorithms are divided into two categories according to two-stage and one-stage implementation. At the same time, the reference whether to use anchor frame is divided into two ways based on anchor frame and non-anchor frame. The realization principle of the algorithm is discussed in detail around the development of a more mature detection system based on anchor frame, and the difficulties and key technologies facing the current object detection system are pointed out. Finally, the development direction of object detection algorithm is prospected.