Software and Hardware Co-Design Method for Ship Detection in Remote Sensing Images

The on-board processing of ship target detection in remote sensing images can transmit the detection results directly back to the ground and obtain fast processing speed. However, the increase in the amount of remote sensing image data leads to the decrease in the timeliness of on-board remote sensing image ship detection. In this article, a software and hardware co-design method for target detection in remote sensing images is proposed. In this method, the processing system in Zynq system on chip (SoC) is used for pre-processing as well as integration and output of detection results, and the programmable logic in Zynq SoC is used for accelerating the MobileNetV3 target detection method. The experimental results show the proposed method can accomplish satisfying ship detection of remote sensing images. Meanwhile, the average processing time of a remote sensing image of 64 * 64 size is 2.59 ms, which is a relatively fast processing speed.