Developed a computer vision based parking occupancy detection system using Python and OpenCV to monitor the availability of parking spaces from a live video feed. The application processes each frame in real time, isolates predefined parking regions, and determines whether a space is occupied by analyzing pixel intensity after image preprocessing.
The detection pipeline combines these steps to reduce environmental noise and improve system robustness under varying lighting conditions.
Parking slots are mapped using manually defined pixel coordinates. The system extracts each slot region, calculates the number of non-zero (white) pixels in the thresholded slot block, and compares this value to a calibrated threshold. If the density of thresholded pixels exceeds the limit, the space is classified as occupied; otherwise, it is flagged as vacant.
The user interface overlays live occupancy status (red bounding boxes for occupied slots, green for vacant) directly onto the video stream and displays the total number of available spaces dynamically in real time.
The project focuses on efficient image processing techniques, real-time analysis, and practical deployment of computer vision for intelligent parking management. It strengthened my understanding of multiple core concepts: