Road Vehicle Detection Based on Feature Fusion Between Frames

With the rapid economic development, motor vehicles are becoming more popular, and artificial intelligence applications on the road are emerging in endlessly. In current road vehicle detection algorithms, most of them use single-frame image information intercepted from video sequences for vehicle detection. This method does not take into account that the difference between frames in the video sequence is mainly the motion background information. Aiming at this design limitation, this paper proposes a target detection method based on IFFF (Inter-Frame Feature Fusion). In the input part of the model, in addition to adding the picture of the current frame, the feature map output of the previous frame will be added to enrich the information of the current frame and improve the detection performance of the current frame. At the same time, a spatial pyramid pooling structure is added to the network to further integrate local and global features to improve the ability to detect vehicles. Experiments show that the method proposed in this paper can improve the detection effect of vehicles in road scenes. Compared with the original CenterNet detection network, the mAP index is improved by 4.3%.