Bayesian Hierarchical Detection Framework for Parking Space Detection
People
Ching-Chun Huang
Concept of the proposed vacant space detection process.
Abstract
Using intelligent surveillance systems to monitor parking lots is becoming a practical application. To assist users in efficiently finding empty parking spaces, a parking lot management system may perform parking space detection to identify available parking spaces over time. However, maintaining such information manually needs lots of human resource. Therefore, automatic parking space detection has been employed in many systems for counting the number of available parking spaces, identifying their locations and monitoring changes of their status over time. Three-day tested sequences, detection results and comparisons are provided for references.
Below, we list the detection results based on proposed method. The images from top-left to bottom-rihgt are the tested image, the shadow region labeling, the expected shadow map, the final vacant space detection, the car region labeling, and the expected car map. In the foruth image, the final vacant space detection, the red rectangle represents the ground truth of vacant spaces and the green rectangle represents the detection results by the proposed method.


Below, we compare the vacant space detection results of proposed method with other three methods. From left to right, the images are the detection results based on the proposed method, the Huang's method [1], .the Qi's method [2], and the Dans' method [3],


Recently, we test our system in another parking lot. Below, the complete parking space detection results in this new parking lot are provided for further reference. The images from top-left to bottom-rihgt are the tested image, the shadow region labeling, the expected shadow map, the final vacant space detection, the car region labeling, and the expected car map.

  • 1.Video sequence of detection results in the new parking lot

  • For the research reproduction and the further comparisons, three-day tested sequences with ground truth and associated annotations are available for the research purpose. The datasets include testing images, the ground truth of parking statuses, the camera calibration matrix, and the space layout of the parking lot. Note that the ownership of the datasets is belonged to Industrial Technology Research Institude (ITRI) of Taiwan and for research purpose only. Please acknowledge ITRI and cite the relative references if you use the datasets. Download the datasets from here. [Download Datasets]
Publication
Ching-Chun Huang, and Sheng-Jyh Wang "A Hierarchical Bayesian Generation Framework for Vacant Parking Space Detection", IEEE Transactions on Circuits and Systems for Video Technology, vol. 20, no. 12, pp. 1770-1785, Dec., 2010. (SCI, EI) NSC 97-2221-E-009-132.