Real-Time Object Detection in Underground Parking Lots using PointPillars

In this paper, we address the problem of object detection in the complex environment of underground parking lots by proposing a real-time object detection system based on PointPillars. This system leverages the PointPillars model to achieve fast and accurate detection of vehicles and pedestrians, coupled with the real-time data processing capabilities of ROS. Our system successfully implements real-time object detection in underground parking lots. Experimental results show that the system achieves a detection accuracy of over 97% on the test set, demonstrating excellent real-time performance and accuracy in practical applications. This study not only enhances the safety and management efficiency of underground parking lots but also provides a new solution for object detection technology in similar scenarios.