Vehicle density
Lane-by-lane counts, straight from video.
Point it at an intersection feed and it reports how heavy each lane is, what is in it and what is blocking it.
What goes in
A video stream, an uploaded clip or a mock feed of an intersection. Frames are sampled and sent to Gemini with a structured-output prompt, so the answer comes back as data, not a paragraph.
What comes out for every lane
- Density level (low, medium or high) with an approximate count range
- Vehicle mix: jeepneys, buses, cars, motorcycles, motorelas and others
- Queue length
- Blockage type, such as loading in the lane or a stalled vehicle
- Standing water
- A confidence level
People, not just vehicles
A jeepney carries more people than a car. We plan to weight lane priority by estimated passengers instead of vehicle count alone, and we will publish the occupancy assumptions we use.
How accurate is it?
Counts are shown as ranges, never as exact numbers. We measure them against manual counts from the same footage and report accuracy by condition, because night and rain are harder than a clear afternoon. When confidence is low, the system falls back to the existing fixed-time plan.
Faces and license plates are blurred, and nothing here identifies a person or a vehicle. See safety and privacy.
Keep reading
See it for yourself.
The app runs on sample data. Open the demo account and try the dashboard, the sandbox and the assistant.