Vehicle target detection dataset

The public UA-DETRAC data set was used to train the vehicle target detection model in this study. This data set was captured predominantly in the Beijing-Tianjin-Hebei area. Using a Canon EOS 550 camera, 10 hours of video were captured at 25 frames per second, and then the footage was processed. Each frame had a resolution of 960*540 pixels. The UA-DETRAC data set contained 8,259 manually labeled vehicles, with a total of 1,210,000 labeled vehicles. The labeled types included cars, buses, trucks, etc. The data set used contains a total of 140,131 images, of which 70% of the data set is used for training, 15% is used for validation, and 15% is used for testing.