The EPIC iO DeepInsights video analytics produces events from each rule configured for an ROI. Some ROIs will have more than one rule, as shown in the driver area configuration we set up during testing. The DeepInsights Portal includes a feature to monitor a real-time view of the generated telemetry to validate rule accuracy. The following screenshot shows the debugger console view accessed using a separate tab on the rule definition UI.

Tracklets indicate how often the rule detected a configured condition or class of object. We used this feature to validate that the rules we configured produced the results we expected from our sample video loops. For our testing, the telemetry generated from rules in the Scene Editor was forwarded as Alerts to the DeepInsight Public Cloud Service.

The following screenshot shows the browser/web UI for reviewing events collected from all connected buses in a single console host on the DeepInsights Public Cloud Service.

Our testing also used computer simulation to produce time series of data similar to what would be produced by IoT sensors positioned on the bus. We simulated bus speed, inside air temperature, particulate concentrations, oxygen levels, and fuel tank capacity. The following screenshot shows a dashboard with four panels displaying near real-time conditions on a simulated bus. Notice that some of the data on the dashboard come from environmental sensors and an aggregated total occupancy derived from computer vision analytics.

We also used a stream of actual GPS coordinates from a typical urban bus route to stream near real-time locations from our lab to the DeepInsights Public Cloud Service to test and demonstrate the mapping capability of the cloud dashboarding tools. The results are captured in the following screenshot.