Underground Parking Lot Empty Slot Detection Dataset

With the acceleration of urbanization, the management of underground parking lots has become increasingly complex, with low space utilization and traffic congestion being common problems. Existing manual management and traditional surveillance cameras cannot monitor empty slots in real-time and efficiently, incurring high costs and with limited accuracy. This dataset aims to assist in identifying and detecting empty slots in underground parking lots using deep learning technology, improving parking management efficiency and user experience. Images are collected from high-definition cameras installed on the ceiling of parking lots, covering various weather and lighting conditions. Quality control involves multiple rounds of manual annotation and consistency checks to ensure the consistency and accuracy of each annotation. The annotation team consists of 12 experts with backgrounds in traffic management and computer vision. Preprocessing stages include data cleaning, denoising, and normalization, using the latest image enhancement techniques to augment the samples. Data is stored in JPG format, organized into folders classified by time and location.