A dated machine count for each yard you track, each month.
Per-site counts of heavy machinery, read from very-high-resolution imagery, for the model your market share is built on.
Spec
- GSD
- VHR, ≤0.5 m
- Cadence
- Monthly
- Platform
- VHR satellite or aerial
- Spectral
- RGB
- Unit
- one dated count per site per month
- Thermal
- none
01What you get
The files
| File | Format | What's in it | Cadence |
|---|---|---|---|
| Monthly machine count | CSV | One row per site per month: site ID, image date, count of machines standing. Lands in your warehouse or spreadsheet as it is, keyed to your own site IDs. | Each month |
| Segmentation overlay | GeoPackage | Machine footprints drawn on the source image for each pass. Open it in your GIS next to the count when a number looks wrong and you want to see which machines were counted. | Each pass |
| Site-over-time table | CSV | Count history per site, month by month, with gaps left where a pass was unusable. Built for trend lines and for the monthly competitor pack. | Each month |
| Coverage note | For each pass, which sites were counted, which were skipped for cloud and which have areas under cover or out of frame. | Each pass |
File by file
- File
- Monthly machine count
- Format
- CSV
- What's in it
- One row per site per month: site ID, image date, count of machines standing. Lands in your warehouse or spreadsheet as it is, keyed to your own site IDs.
- Cadence
- Each month
- File
- Segmentation overlay
- Format
- GeoPackage
- What's in it
- Machine footprints drawn on the source image for each pass. Open it in your GIS next to the count when a number looks wrong and you want to see which machines were counted.
- Cadence
- Each pass
- File
- Site-over-time table
- Format
- CSV
- What's in it
- Count history per site, month by month, with gaps left where a pass was unusable. Built for trend lines and for the monthly competitor pack.
- Cadence
- Each month
- File
- Coverage note
- Format
- What's in it
- For each pass, which sites were counted, which were skipped for cloud and which have areas under cover or out of frame.
- Cadence
- Each pass
02Pipeline
How it works
01 You supply
You send the site list
Yards, pits or terminals as polygons (
GeoJSON,shapefileorGeoPackage) with your own site IDs. Start with the dealer and competitor yards that matter most to your share figure.02 We do
We pull the monthly pass
The latest usable very-high-resolution image of each site is taken. A cloudy date moves that site to the next clear window.
03 We do
Machines are detected and outlined
Standing machines are found and outlined in the image. No make or model is read.
04 You get
You get the dated count
Count, overlay and history table per site, each row tagged with the image date it came from. The counts load into your warehouse or model the same way each month, so last month's figure never gets restated quietly.
03Limits
Where the count stops
| Condition | Effect | Note |
|---|---|---|
| Machines under cover | A machine inside a shed, under a canopy or beneath a tarpaulin does not show. | The coverage note flags sites with covered areas. |
| Tight stacking | Machines parked close together or partly hidden by other equipment can merge or be missed. | The count is of what can be seen from above. |
| Small machines | At ≤0.5 m a pixel is up to half a metre across. A large excavator spans many pixels. | A compact loader or skid steer spans few, and is counted with less certainty. |
| Cloud and the pass date | Each count carries the date of its image, not a fixed day of the month. | A cloudy pass delays that site to the next clear window, and the gap stays visible in the history. |
| Standing, not working | A still image shows a machine standing there. | It says nothing about whether it was working, idle or parked for the season. |
| Class, not make | Machines are counted by body type where that is separable. | Make and model are not read. |
Operating envelope
- Condition
- Machines under cover
- Effect
- A machine inside a shed, under a canopy or beneath a tarpaulin does not show.
- Note
- The coverage note flags sites with covered areas.
- Condition
- Tight stacking
- Effect
- Machines parked close together or partly hidden by other equipment can merge or be missed.
- Note
- The count is of what can be seen from above.
- Condition
- Small machines
- Effect
- At ≤0.5 m a pixel is up to half a metre across. A large excavator spans many pixels.
- Note
- A compact loader or skid steer spans few, and is counted with less certainty.
- Condition
- Cloud and the pass date
- Effect
- Each count carries the date of its image, not a fixed day of the month.
- Note
- A cloudy pass delays that site to the next clear window, and the gap stays visible in the history.
- Condition
- Standing, not working
- Effect
- A still image shows a machine standing there.
- Note
- It says nothing about whether it was working, idle or parked for the season.
- Condition
- Class, not make
- Effect
- Machines are counted by body type where that is separable.
- Note
- Make and model are not read.
04The job today
Why counts come from dealer calls
You need to know how many machines stand in a competitor's yard this month, and whether that is up on last month. The number arrives from a dealer call, a regional manager's impression, or registration figures that land late and in aggregate.
Each has a different date and a different idea of what "on hand" means, and none is per site. Reconciling them by hand is slow, and the gaps get filled with a judgement call. When the share model is rebuilt, the per-site unit underneath is an estimate somebody has to defend.

05Fit
Who it's for
+ For
- Market intelligence analysts at equipment manufacturers who track stock across dealer and competitor yards.
- Teams that want a dated per-site count as the input to a share, utilisation or replacement-cycle model.
- Analysts who will reconcile the count against a dealer's own stock list.
−Who it's not for
- Teams that need make or model identification. This counts machines.
- Utilisation or activity reads. A still image cannot tell idle from working.
- Live or daily fleet tracking. The cadence is monthly.
- Fleet telematics replacement for your own machines. That data comes from the machines.
06FAQ
Questions from the evaluator
Does it identify make or model?
No. It counts machines standing on site, by body type where that is separable.
What does ≤0.5 m resolution miss?
Machines under cover, machines hidden by others in a tight stack, and small compact equipment that spans only a few pixels.
How current is the count?
As current as the latest usable pass for each site. Every row carries its image date.
What happens on a cloudy pass date?
That site waits for the next clear window. The history table shows the gap rather than filling it.
Can it tell idle from working machines?
No. It is a standing count from a still image.
How do we check it against our own data?
Every count comes with the overlay, so you can set it beside a dealer's stock list or a yard visit for the same week. Tell us where your ground truth sits when the pilot is scoped.
Who owns the files, and what about the imagery?
Use terms for the delivered files are written into the contract before the pilot starts. Imagery suppliers keep their own licences.
How is a pilot priced?
Scoped on its own: number of sites, how many months, and whether you want a history back-filled. You get a written scope to take to procurement.