Using Convolutional Neural Networks To Improve Road Safety And Save Lives

Lois Anne Leal and Chaitree Sham Baradkar participated in the innovative iRAP Omdena 8 week Challenge earlier this year to find AI-based solutions to map crash risk and generate iRAP star rating attribute data. They were just 2 of 50 AI and machine learning engineers crowd-sourced from around the world focussed on finding a solution for the accelerated and intelligent collection and coding of road attribute data (AiRAP) to save lives.

Lois is a Science Research Specialist, Machine Learning Engineer and Certified TensorFlow Developer from the Philippines and Chaitree is a Data Scientist at PharmaACE and Machine Learning Engineer at Omdena from Pune, India.

The iRAP Omdena Challenge had 3 main objectives:

• Source geo-located crash data and produce iRAP Risk Maps of the historical crashes per kilometer and crashes per kilometer traveled for each road user.
• Source road attribute, traffic flow, and speed data to the iRAP global standard and map the safety performance and Star Rating of more than 100 million km of road worldwide.
• Produce repeatable road infrastructure key performance indicators that can form the basis of annual performance tracking.

Lois and Chaitree contributed to the second objective by automatically sourcing the crucial component of vehicle count under the traffic or vehicle flow attribute using satellite imageries with the help of Artificial Intelligence.

Read their published blog: “Using Convolutional Neural Networks To Improve Road Safety And Save Lives“.