Method and device for reducing ADAS target detection false detection and medium

Abstract

The invention discloses a method and device for reducing ADAS target detection false detection and a medium, and the method comprises the steps: carrying out the data matching of all target detection frame data detected by a current frame image and all target detection frame data detected by a previous frame image; if the data of the current frame of target detection frame data and the previous frame of target detection frame data are within a preset data matching threshold value, performing corresponding marking, performing prediction according to the matched change value of the current frame of target detection frame data, predicting the target detection frame data of the previous frame of target detection frame data appearing in the current frame of image, and outputting the predicted target detection frame data. The prediction data is corrected according to the matched current frame target detection frame data, the previous frame target detection frame data is replaced and updated, and the actual target detection frame data of the current frame is subjected to target filtering and screening; and if the target detection frame data in a plurality of continuous frames cannot be matched with the replaced and updated target detection frame data for a plurality of times, judging that the target is a false detection target.

Description

technical field [0001] The invention relates to the technical field of target detection, in particular to a method, device and medium for reducing false detection of ADAS target detection. Background technique [0002] With the development of artificial intelligence technology, object detection technology has been widely used in Advanced Driving Assistance System (ADAS). By detecting the targets in the vehicle monitoring video, it can provide reference, reminder and collision warning for the driver, and improve the driving safety. [0003] The current method for realizing target detection generally uses a target detection technology based on a neural network, which mainly detects a specified target based on image data. However, due to technical limitations, it is difficult to achieve complete accuracy in practical applications, especially for complex scenarios without prior knowledge, there are often missed detections and false detections. [0004] To improve the detection...

AI Technical Summary

Problems solved by technology

These improved methods mainly improve the image processing algorithm of a single image frame by frame. Because these methods isolate the data of each frame of the video sequence, there is no reference to the detection results of the previous and subsequent time periods, and the timing characteristics of the video data are not fully utilized. When the detection accuracy reaches a certain level, it is difficult to further improve, and the phenomenon of target misdetection is relatively large

Images