Research on Vehicle Parking Position Detection Based on Swin Transformer Semantic Segmentation

Abstract

Determine whether the vehicle is parked in the specified area, which has high application value in industry, transportation, parking lot, etc. Aiming at the problems of rough results and high maintenance costs, a vehicle parking position detection method based on Swing Transformer semantic segmentation is proposed. The vehicle semantic results obtained by Swin Transformer semantic segmentation algorithm are taken as the main features of vehicles in the picture. Canny algorithm is used to obtain vehicle contour to improve detection accuracy. Calculate the relationship between the vehicle contour and the hand drawn warning line, and compare with the threshold value to determine whether the vehicle is parked in the specified area. Through simulation, industrial application and road application, the method can realize the normative detection of vehicle parking position.