Intel is in talks with startups and academic researchers who are capable of using the dataset to build a technology that can help drivers react to a variety of potential hazards in a split-second.
Bad roads making you wary about stepping out? Intel has a solution.
The US chipmaker is working on a software that will alert drivers about road conditions – including potholes and the surrounding environment – well in advance.
The company – which has created a database of roads in Hyderabad and Bengaluru – believes the rate of accidents can be brought down through a proper understanding of the road.
Intel, with help from the International Institute of Information Technology (IIIT), Hyderabad has created a dataset that detects people, vehicles, animals, vegetation, immovable structures, traffic signs etc. on the road.
The company is also in talks with startups and academic researchers who are capable of using the dataset to build a technology that can help drivers react to a variety of potential hazards in a split-second.
“About 17 die of road accidents in an hour. And, most of them are young men. If the computer can solve problems like this – any amount of money spent on this project is worth it,” Nivruti Rai, Country Head of Intel told ET about its smart mobility project.
AI and vision technologies can be leveraged to give drivers a few extra seconds to avoid collisions.
Intel’s intent, Rai said, is to solve national-level problems. “We are keeping this data available as open-source and any startup or academia can access it and build technology that can avoid collision or accidents. Intel alone cannot do it. I need data. I need algorithms. So, I have to partner with academia. We are partnering with IIIT Hyderabad and IISc in Bengaluru for AI-based solutions,” she said.
The company hopes the dataset, consisting of over 50,000 image frames and 10,000 segmentation annotations, takes the shape of algorithms that could go into the cars.
“Some car manufacturers are already doing similar work. But, we want to build a solution that fits into old cars too and not necessarily the new ones,” Nivruti said.
The India Driving Dataset also facilitates autonomous vehicle training and advances AI research for smarter and safer mobility. The dataset is the result of a year-long collaboration with IIIT.
The aim is to unleash community-driven innovation through this initiative.
“We will use AI tasks to increase the amount and diversity of different types of roads/region,” said Professor CV Jawahar, Dean of R&D at IIITH.
The institute is also looking at areas such as specific driver assistance and road safety issues that can be enabled by such datasets to understand driver behaviour and the quality of traffic infrastructure.
It is also reaching out to students to make use of the datasets.
“We, along with the International Conference on Computer Vision (ICCV), had hosted a challenge and similar efforts are being planned for Indian students and startups in December. We are working with industry and startups to convert research into technology,” Jawahar said.
Date: November 22, 2019