The International Institute of Information Technology (IIIT) in Hyderabad and Intel worked together for over a year to create the world’s first open dataset of driving conditions in India so that an autonomous vehicle training can be initiated for the purpose of more road safety and safety to drivers in general. Testing for autonomous vehicles has already started in the U.S states like Arizona, California, Ohio, and Michigan and the availability of the autonomous car services is based on the environment in which the algorithm has been trained and tested.
However, the condition of the roads in the United States and the condition of the roads in India are very different from each other and as such, Indian drivers face twice the problems as the U.S drivers and Indian drivers need to think quick on their feet to overcome the obstacles faced while driving. Needless to say, algorithm regarding the autonomous vehicles would not work if it were to be applied in India with its eccentric driving conditions. The dataset developed by the joint collaboration of IIIT and Intel is based on 50,000 images frames at 1080p and 720p resolution which was taken from a camera attached to a car driven around Hyderabad and Bangalore.
These pictures include 10,000 segmentation annotations with marked drivable and non-drivable areas pinpointing vehicles and objects. During the international launch of the dataset, it has come to everyone’s notice that the dataset has piqued the interest of researchers worldwide. The dataset with its unique elements will hopefully give the nudge needed for the self-driving algorithms to thrive.
Europe is developing their own self-driving algorithm suited for its own cities and China is doing the same. This is only baby-steps which India is taking when it comes to a fully developed self-driving algorithm. India will obviously need several more years before autonomous driving is fully deployed. However, driver assistance is likely to come first to the U.S for commercial purposes which will be closely followed by individual users.