Rapid target visual detection method assisted by millimeter wave radar in vehicle-mounted ADB headlamp

The invention provides a millimeter wave radar-assisted rapid target visual detection method in a vehicle-mounted ADB headlamp, and the method comprises the specific steps: S1, carrying out the data collection of the condition of a pedestrian in front of a vehicle through a camera, and carrying out the processing of the data through an optimized YOLOv5 neural network, and obtaining the key features of image information; s2, radar data is preprocessed, and radar dynamic target capture is realized on preprocessed radar information by applying a Kalman filtering algorithm based on uniformly accelerated motion model extension; and S3, synchronizing radar and camera data through sensor time and space fusion, assisting in rapidly positioning an image area where a target is located by means of radar data projection, completing sensor target matching, realizing rapid and accurate detection of a night moving target by using a data fusion decision-making method, and outputting target detection multi-modal information.