Most fixed-camera footage, in common formats like MP4. The higher and steadier the camera, the better the counts. Wide shots from very far away, heavy rain on the lens and night footage without street lighting are harder.

It depends on the camera angle, resolution, lighting and weather, so we’d rather show you than quote a number. Run a pilot against one of your manual counts, and check any count you like in the annotated video.

How long does it take?

It depends on how much footage there is and its resolution. You can watch the analysis in the panel, and the results are there as soon as it finishes.

Do you work outside Canada?

Yes. Footage from anywhere works the same way.

We count what’s in your traffic video.

Send us footage from a signal camera, a CCTV camera or a GoPro on a pole. You get counts of pedestrians, cyclists, cars, trucks, buses and motorcycles, by direction and in 15-minute intervals, plus a copy of the video with every road user boxed so you can check our work.

Email us about a pilot

Ordvald is new and based in Windsor, Ontario. We’re looking for a few cities to run pilot studies with.

Annotated video frame looking down on a busy crosswalk. Each pedestrian, cyclist and car has a labelled box and a trail showing its path, with a running count in the corner. — A frame of Ordvald’s annotated output. The panel in the corner is the running count.Footage: eQzit / Pexels

What a study gives you

Road users: Pedestrians, cyclists, cars, trucks, buses and motorcycles. Each one is tracked through the scene and counted once, so someone waiting at the corner for two minutes is one pedestrian, not two hundred.
Zones: You draw the crosswalks, sidewalks and roadway on a frame of your video. Each zone keeps its own counts.
Direction: Northbound, southbound, eastbound and westbound, once you mark which way is north.
Time: 15-minute intervals, the peak hour and the peak hour factor.
Crossings: Pedestrians who cross the roadway away from a marked crosswalk are flagged.
Files: A counts spreadsheet (CSV), an event log with one row per movement, and the annotated video.
Not yet: Turning movement counts, PDF reports and e-scooters. We’re working on them.

How a study works

1.You record. Any fixed camera works if it looks down on the area and doesn’t move. If the site has no camera, a GoPro on a pole for a day is enough.
2.You upload the video from your browser, or send it to us.
3.You mark the zones and set north, or we do it for you. It takes a few minutes per site.
4.The analysis runs. You can watch it frame by frame while it works.
5.You download the counts, the event log and the annotated video.

Annotated street-level video frame of a Montreal intersection with pedestrians, two cyclists and parked cars, each in a labelled box. — Street-level footage works too. A Montreal intersection with two cyclists crossing.Footage: German Korb / Pexels

Checking the numbers

A count is only useful if people believe it, so every study comes with the evidence. The annotated video shows each road user with a box, a label and the path it took. The event log lists every counted movement with its time, zone and direction.

If a number looks wrong, you can find the moment it happened and watch it.

events.csv

Example rows. The columns are the real export format.

date,time,video_time,interval_15min,location,site_id,zone,zone_type,event,road_user,direction,outside_crosswalk,track_id2026-06-16,17:02:14,00:02:14,17:00,Main St & 1st Ave,S-014,North crosswalk,crosswalk,passage,pedestrian,EB,no,4122026-06-16,17:02:17,00:02:17,17:00,Main St & 1st Ave,S-014,Main St roadway,road,passage,cyclist,NB,no,4152026-06-16,17:02:19,00:02:19,17:00,Main St & 1st Ave,S-014,Main St roadway,road,passage,bus,WB,no,4092026-06-16,17:02:21,00:02:21,17:00,Main St & 1st Ave,S-014,Main St roadway,road,outside_crosswalk,pedestrian,,yes,4182026-06-16,17:02:26,00:02:26,17:00,Main St & 1st Ave,S-014,Main St roadway,road,passage,car,SB,no,420

Privacy

Ordvald detects road users by type. It doesn’t recognize faces, read licence plates or follow anyone from one video to the next.

Footage is uploaded over an encrypted connection, processed in Canada, and used only to produce your counts.

Pilots and pricing

We price per study, by the hours of video. There’s no hardware to buy and nothing to install.

The fairest test of Ordvald is a count you already trust. Send us video from a site you’ve counted by hand, and compare our numbers with yours.

Questions

What video works?: Most fixed-camera footage, in common formats like MP4. The higher and steadier the camera, the better the counts. Wide shots from very far away, heavy rain on the lens and night footage without street lighting are harder.
How accurate is it?: It depends on the camera angle, resolution, lighting and weather, so we’d rather show you than quote a number. Run a pilot against one of your manual counts, and check any count you like in the annotated video.
How long does it take?: It depends on how much footage there is and its resolution. You can watch the analysis in the panel, and the results are there as soon as it finishes.
Do you work outside Canada?: Yes. Footage from anywhere works the same way.

About Ordvald

Ordvald started in Windsor, Ontario, as a way to get pedestrian and cyclist counts out of traffic-camera footage without sending staff out with clickers.

It’s for municipal traffic and transportation teams, transportation researchers at universities, and the engineering firms that run count studies for cities, in Canada or anywhere else.

The name joins two Old Norse words: ord, the record, and vald, power.

Start a pilot

Email us. Tell us where the camera is, how many hours of video you have and what you need counted.

Email us about a pilot hello@ordvald.ca