Count Objects
On-deviceGet a first count automatically, then tap to fix it — done entirely on your device.
This is an aid, not a certified count — always check the number before relying on it.
Detect objects in a photo automatically, then tap to add or remove markers until the count is right. Adjust sensitivity and object size with visual handles, switch between dark-on-light and light-on-dark, and export an annotated image or a CSV of marker positions. An on-device engine (~4 MB) powers the automatic pass; nothing is uploaded.
About this tool
What you provide
- A photo in any format your device can decode — JPEG, PNG, WebP, GIF or AVIF.
- Sensitivity and Largest object, shown as visual size handles rather than pixel numbers.
- A dark-on-light / light-on-dark toggle for how objects sit against their background.
What you get
- A live count badge, updated as markers are added or removed.
- An annotated PNG with numbered markers drawn on the photo.
- A CSV with one row per marker (id, image x/y, automatic or manual).
Limits
- Automatic detection needs a one-time ~4 MB engine download; manual marking works without it.
- The automatic pass is a first draft — overlapping or unevenly lit objects need manual correction.
- Very large photos are downscaled for the detection pass only; markers and the exported image stay at full resolution.
How to use it
-
Choose a photo
Pick a photo from your device — it's read locally and never leaves your device.
-
Download the detection engine
The first time, tap the one-time ~4 MB download to enable automatic detection. It runs entirely on your device.
-
Run automatic detection
Tap Detect automatically to get a first-pass set of markers, adjusting Sensitivity, Largest object and the background toggle as needed.
-
Correct the count by hand
Tap an empty spot to add a marker or tap an existing one to remove it — the badge always shows your current total.
-
Export the result
Download the annotated photo or a CSV of marker positions to keep or share the count.
Frequently asked questions
Does my photo get uploaded?
No. The photo is read, marked and counted on your device; nothing is sent to a server. Automatic detection needs a small one-time engine download (about 4 MB), after which it works offline.
How accurate is the automatic count?
It's a first draft, not a certified count — lighting, overlapping objects and clutter can trip it up. That's why manual correction (tapping to add or remove a marker) is the main way to use this tool, and the badge always reflects your edits, not just the automatic pass.
What do Sensitivity and Largest object control?
Sensitivity sets how small or faint an object can be and still get picked up. Largest object caps how big a single detected item can be, so a big background patch is never counted as one giant object. Both show as a growing or shrinking dot rather than a pixel number.
What does the exported CSV contain?
One row per marker, with its id, image coordinates and whether it was placed automatically or by hand — useful for checking a count or re-importing it elsewhere.