Millions of deaths occur from traffic accidents every year with many efforts trying to reduce the frequency and severity of these traffic accidents. The most effective way to tackle this problem is by means of an extensive program of road safety management, in which road safety modelling plays an essential part.
The modelling process attempts to adjust a model to the crash data, the geometric and operational characteristics of the road, and the environmental conditions, incorporating the most important factors. Numerous modelling techniques have been developed to improve the representation of reality, allowing for the employment of techniques that are more appropriate to the problems.
There have been limitations to these approaches, allowing for new opportunities, such as machine learning that SANRAL is currently exploring. This allows to improve road safety, reduce congestion and information infrastructure development.
Machine learning can be used to detect and segment objects within the camera frame. These objects can then be classified based on pre-trained image classification. Which ultimately allows for the detection and classification of different types of vehicles, pedestrians, different types of animals, cyclists, etc. The possibilities are infinite, based on the data available.
Currently, there is ample data on the above-mentioned classification types. While this is still in the exploratory phase within South Africa, it does come with significant risk and efforts are being made to understand how to effectively use this technology, while maintaining strict compliance with legislation as it relates to the privacy of the road users.