Kolkata City Police Plans To Use AI for Identifying And Prosecuting Helmetless Riders And No-Parking Rule Violators

Over 2,500 CCTV cameras will use artificial intelligence in Kolkata, 125 of which can detect helmetless bikers and illegal parking in real time, prosecuting offenders. The Kolkata Police Department has requested a detailed presentation from technology specialists in December to help determine the latest machine learning techniques increasingly used to detect helmetless bicyclers and illegal parking.

7.3 lakh traffic violations were discovered on Kolkata streets in 2020, out of the 39.4 lakh documented by CCTV data. Academic bodies are now undertaking several projects to identify helmetless bikers and their number plates in real-time. The thrust on technology will also reduce the risk to officers who physically try to stop these bikes at personal risk. The police also want software models to identify illegal parking and trace the vehicle licence plates.

This is being achieved by adding a detection card to CCTV cameras, which helps detect helmetless bikers by sending out an alert. When cameras detect vehicles in no-parking zones, alerts will be sent out. By adding a webcam or a CCTV as input, this application can be made in real-time. IIT Hyderabad has developed a program for automatically detecting bikers without helmets in surveillance videos.

Once the cops zero in on the project, a detailed proposal will be sent to the state. Arijit Sinha, DCP of Traffic, said, “There are some plans in this regard. It is too early to comment till anything is finalised.” 

For illegal parking, the system uses an IP camera to continuously capture and transmit image sequences to a base station, where CNN is used to perform automatic image extraction when detecting illegally parked vehicles. The proposed application will be robust against environmental influences, such as shadow effects, light changes, and different weather conditions. Both systems will produce challenges. It will be a trial system with a web interface to check the alerts.