Open data

NYC traffic-camera vehicle counts — open dataset

New York City publishes 917 live street cameras but publishes no counts. Bilbo Data runs computer vision across that network and writes down what passes: cars, trucks, buses and motorcycles, per camera, per timestamp, with a brightness and lit/unlit flag so you can separate day from night. It is free, it needs no login, and it is the same file the site itself reads.

Coverage

29 cameras counted, of 917 mapped

Rows

37,207 and growing

Fields

timestamp, camera id, name, car, moto, bus, truck, total, brightness, lit, samples, stale

Formats

CSV (rolling) and Parquet (daily)

Cadence

continuous; the counter commits straight to the repo

Licence

Free to use with attribution to Bilbo Data

30-day average

90.1 vehicles in frame

30-day peak

237 vehicles in frame

What is in it, honestly

These are counts of objects in a frame, not a census. The cameras are low-resolution and many are dark at night, which the brightness and lit columns let you filter on. The stale column flags frames the DOT feed repeated instead of refreshing. We publish the flaws with the data rather than smoothing them away — see the about page for where the ceiling actually is.

What people use it for

Traffic-volume comparisons between intersections, before-and-after studies around street redesigns, congestion-pricing effects, bus-lane utilisation, truck-route enforcement questions, and time-of-day demand curves for anyone modelling deliveries in the city. If you want a specific intersection, every camera has its own page under the camera index.