ISP Parameter Optimization and FPGA Implementation for Object Detection in Low-Light Conditions

Camera applications using object detection, such as automated driving and surveillance cameras, are becoming more common. It is necessary to detect objects near sensors with low latency in environments with low visibility such as nighttime conditions. In this paper, we have optimized the parameters of an Image Signal Processor (ISP) to maximize the accuracy of object detection in low-light conditions. The object detection accuracy was improved from 66.7% to 80.4% by optimizing ISP parameters using a Genetic Algorithm (GA). The optimized Image Signal Processor (ISP) was implemented on an FPGA and confirmed to operate at 90 MHz.