Objects and features, detection

Q: What is the accuracy street widths and length?

The accuracy of our street width measurements is within approximately 1 foot. Your question regarding length accuracy requires clarification. We are not trying to measure exact road length for today, but instead we have an area for each segment.

Q: Signs – can you detect reflectivity and flag deficient signs? If so, is this available ‘out of the box’ or a solution that will need to be developed?

Road markings – are you able to detect reflectivity and flag deficient markings? If so, is this available ‘out of the box’ or a solution that will need to be developed?

A: As of today, explicit reflectivity modeling as well as deficient signs, and road markings are in our roadmap. Alternatively, it requires 1-2 months for the ML model preparation and development.

Q: Do you do any AI around tree inventory and species identification for trees adjacent to roadways?

A: We have built an extensible modular architecture that can help inventory many objects. Certainly, we detect and classify trees among other classes. Roadly can provide trees cloud as a separate layer.

Q: Water asset conditions – Ie tracking raised/damaged manholes/drains/catch basins?

A: Like trees, we detect and classify manholes, drains, and catch basins among other classes. However, their condition is not recognized at this time. Roadly can show all unique manhole images on a separate page of the platform as a tool for visual estimation of their condition.

Image. Semantic point cloud