11/25/2020 09:14 GMT + 7
The group of students in Ho Chi Minh City proposed the idea of using AI to divert and handle traffic violations in the area.
Pham Thi Thien Thao (Ho Chi Minh City Academy of Cadres) and Le Nguyen Chi An (Ho Chi Minh City University of Technical Education) propose the idea of using artificial intelligence to determine traffic flow, diversion and detect traffic violations.
|The computer recognizes and labels vehicles in traffic through the images captured from the camera. (Photo: group of authors)|
To do this, images captured from traffic cameras will be analyzed in real time by computers, using artificial intelligence and deep learning to predict traffic jams and traffic violations.
The authors believe that traffic jams and congestion in Ho Chi Minh City are largely due to the large traffic volume. Besides, the awareness of the pedestrians is still poor, leading to congestion and accidents. Meanwhile, the traffic police force is not enough to monitor all the time, every corner to detect violations or traffic flow.
In the proposed solution, two students use computer vision technology to calculate the vehicle flow in a given period of time. From there, proposing an alternative route in case of high traffic volume. At the same time, record cases of road users violating the road traffic law.
Specifically, the image from the traffic camera will be detected and identified by the YOLOv4 algorithm with real-time deep learning model. Each vehicle is then labeled and tracked. Illegal conduct will be extracted and saved for processing. Along with that, based on the number of vehicles per unit of time, the algorithm will analyze the risk of traffic jams and propose a new direction of movement, warning traffic flow.
This solution of the student group was awarded the encouragement prize in the young creative contest “Green Transport” 2020.
Answering ICTnews, Thien Thao said, from the idea to the actual implementation will face many difficulties. The amount of images and videos captured from the camera takes up a large amount of data, so it requires a strong infrastructure and computers to process in a very short time.
Besides, the aforementioned data belongs to the management of state agencies, so it is difficult to access. This amount of data is very important for computers to learn and analyze for accurate forecast results.
In addition, video recording equipment must be good enough to accurately recognize vehicles at times of heavy traffic.
The group came up with the idea that they were still in school so they were quite busy with their studies. The lack of capital and other resources also makes this project just stop at the proposed level.
Technology applications to solve urban traffic problems
Many startups around the world are looking for solutions to reduce congestion and improve the face of urban traffic.