The project was to develop a system to detect helmet defaulters in real-time and report the license plate to the concerned authorities. The system had three stages:
- Helmet Detection
- License plate identification
- Real-time processing
Need for the Project
In spite of it being a mandate by the government under the motor rules act, many people chose to avoid wearing helmets while riding a two-wheeler. This leads to many deaths due to damage to the brain in the course of an accident. According to TOI about 43 people die in a day due to them not wearing seat belts or helmets and the survival chances of people wearing a helmet increases by 42 %. To prevent these accidents, there needs to be a system that could detect helmet defaulters beforehand.
- Application can be extended to identifying the stolen vehicle, speeding vehicles, etc
- Reduced number of deaths due to not wearing a helmet
- Real-time detection leads to quick actions by authorities
We used a deep learning model to detect a person wearing a helmet, and applied image processing for extracting number plate. Kafka was used for live video streaming in order to obtain low latency and high throughput.
We successfully implemented the system in real-time and detected the helmet defaulters, further reporting them using the license plate.
In the above result, the biker is detected not wearing a helmet, thus, the license plate will be recorded.
On applying image processing techniques, the license plate of helmet defaulter is recorded and reported to the concerned authorities.
In this output, the biker is wearing helmet, thus the license plate after detection, will not be recorded.
See app demonstratino on YouTube: