Abstract:
In recent years, street parking in prohibited areas has become a social problem, especially
in urban and tourist areas. In addition, because street parking can cause traffic congestion and
accidents, real-time detection is required. The detection of street parking has been previously
implemented on the basis of comparisons of videos recorded by fixed-point cameras. However,
this approach has a limited detection area and low accuracy. To overcome these problems, this
study aims towards a real-time street parking detection system that uses dashboard camera
videos. We propose a machine learning method based on the characteristics of on-street parked
vehicles derived by transforming images into text. The object detection model YOLOv3 was
used to analyze videos. We created a dataset based on the coordinate information of 1765
vehicles and the recording vehicle information. We also created a model using random forest
and logistic regression algorithms and evaluated it using the holdout and stratified 5-fold
validation methods. F-measure values of up to 92% and 89% were obtained for the two types
of model, respectively. These results confirm the effectiveness of the proposed street parking
detection method based on bounding boxes and recording vehicle data.