The whopping 11.04 lakh traffic violations detected by the artificial intelligence-enabled traffic enforcement system in just 16 days roughly amount to a cumulative fine of ₹40-50 crore if at least 60% of the offences are taken into consideration.
If traffic violations continue at this pace, the State can meet the capital expense of the five-year ₹232-crore annuity project in less than six months. The State has to repay capital investment for the project in 20 equal instalments in five years.
But again, the shortage of hands for manning the traffic enforcement system is creating delay in generating e-challans, resulting in piling up of cases, which is expected to raise legal and ethical questions. For instance, of the 11.04 lakh cases detected from June 5 to 20, only 0.98 lakh have been forwarded to the intelligent traffic management system (ITMS) for legal action.
“If this is the situation, the backlog of cases will be around 90 lakh after six months, although we expect a decline in cases in the coming weeks,” says an officer associated with the project.
“The ethical question being raised by the poor management of the system is whether it is right to slap a fine only on a handful of violators, while there is no certainty if others will be punished. Secondly, there will be some legal issues for pursuing the traffic offences detected after a certain period of the incident,” he says.
The accuracy in detecting speed violations by sensor-based fixed and mobile enforcement devices is 99%. So is the case with red light jumping. One of the shortcomings of the system is the margin of error in detecting seatbelt violations. If the passenger wears a black or beige coloured dress, the system may detect it as a traffic violation, for which manual scrutiny is essential.
“This is not the issue of the AI-enabled system implemented here alone. Worldwide, the accuracy of detecting seatbelt violation is around 80%. We can achieve an accuracy of 75-80% in two to three months through mature processing,” he says. The accuracy in detecting helmetless riding is around 78-79%, and other offences more than 95%.
Officials say the 146 personnel appointed to process the images captured by the cameras are insufficient to meet the daily target considering the number of cases. Further, only one enforcement RTO is now assigned to district control rooms to approve traffic violations before forwarding them to the ITMS, which is also hindering the smooth functioning of the system.
The State will have to post additional staff on contract basis to process the images and more Motor Vehicles department staff should be tasked to forward the offences to the ITMS, they say.
