Data Science & Analytics
Further Development of AI Tool for Extraction of Roadside Hazards from Videolog Data and LIDAR
Description
Severe and fatal injuries on rural roads are dispersed across many miles of roadway, leading to a labor intensive process for assessing roadside hazards. RENCI, along with the North Carolina Department of Transportation and UNC’s Highway Safety Research Center are developing an Artificial Intelligence tool to extract roadside safety-related features from NCDOT’s previously collected rural video log data.
RENCI's Role
During a one-year pilot program funded by the U.S Department of Transportation, RENCI proved the ability to use AI computer vision techniques to find safety related features in roadside videolog data. In this phase of the project, now funded by NC DOT, RENCI is enhancing our capabilities to not only find these features, such as telephone poles or guardrails in an image, but to use a combination of AI, image analysis, and LIDAR data fusion to geolocate the features. With these developments, we will be able to use input videolog data to produce an accurate location for each of the features of interest.