Illegal Parking Automatic Recognition Dataset

The current transportation industry faces challenges such as low efficiency and high omission rates in monitoring illegal parking. Traditional monitoring often relies on manual inspections, leading to resource waste and potential hazards. Existing solutions usually cannot adapt to complex urban environments such as lighting changes and obstructions, limiting the accuracy and timeliness of recognition. This dataset aims to provide a high-quality image dataset for illegal parking, solving the technical problems of automatically recognizing illegally parked vehicles to meet the needs of intelligent transportation systems. The dataset is collected by surveillance cameras in multiple cities, organized and annotated using a standardized image processing workflow. The data collection process includes capturing under different lighting and weather conditions to ensure data diversity. We implemented multiple rounds of annotation, consistency checks, and expert reviews as quality control measures to ensure high data quality. Data is stored in JPG format, with metadata corresponding to each image file recorded in a JSON file.