Every year, roads and highways are getting busier as the number of vehicles on roads increases day-by-day. The illegal use of land as a parking lot has also caused problems in many countries. Due to the heavy traffic, road safety and planning has become one of the major concerns.

The current practice of monitoring traffic mainly relies on the use of sensors for continuous data collection or on manual visual traffic study, usually a group of workers measured over several days over certain periods of time, which is labour-intensive and time-consuming. Data and analytics results may not be up-to-date and human error may also occur.

All these concerns have led to the development of smart transportation management system based on artificial intelligence (AI) and computer vision technologies. And intelligent vehicle detection and counting are becoming increasingly important in the field of traffic management.

AI-based Traffic Monitoring Solution

Machine learning technology can help improve traffic monitoring and data analytics in the transportation sector. It enhances detection accuracy and turns the collected data from cameras and sensors into valuable insights. Based on this, Anavision recently developed a real-time vehicle object detection algorithm, an intelligent solution for monitoring traffic flow using AI.

Our solution is capable of processing and analyzing a variety of data types, including images and videos. With the aid of computer vision technology, object detection and classification can help distinguish different types of vehicles, such as bus, private car, taxi, cargo truck, boat, etc., and then determine the traffic density in a given district in a given time that can assist on car flow control and road planning. It can also help to see if there is area that is not supposed to be used for parking (illegal land use).

Vehicles object detection — Figure 1: Vehicle object detection can distinguish different types of vehicles, such as bus, private car, taxi, cargo truck, boat, etc.

illegal land use detection — Figure: Real-time data for visualizing if there is illegal land use

The system will accumulate data continuously and keep training and enhancing our AI-model for precise object detection. And it can be trained to differentiate between events that may cause congestion, such as a car accident or rush hour. In the near future, when the system learns that rush hour happens at certain times, it can perform predictive analysis and use the data to improve traffic flow.