Case Study: GRIDSMART improves intersection safety with CVEDIA ITS solutions

Gridsmart improves intersection AI accuracy from 89% to 95% with CVEDIA

Gridsmart, an intelligent traffic system provider, faced the challenge of training its AI algorithms to accurately detect cyclists and other vulnerable road users at intersections, a process hampered by false positives and negatives. The impracticality and danger of collecting real-world data for every possible scenario led them to seek a solution from CVEDIA and its CVEDIA-RT video analytics service.

CVEDIA implemented its solution by creating highly accurate synthetic data to train Gridsmart's AI models, eliminating the need for risky and lengthy real-world data collection. This approach, optimized for Intel hardware using the OpenVINO toolkit, resulted in a 33% performance improvement for Gridsmart's system and a dramatic increase in accuracy from 89% to 95%, reducing errors by 2.2 times.

Gridsmart