Count things in a photo
Counting five hundred pipe ends by hand is an afternoon; recounting them because you lost track is the whole evening. This counts them in a second, numbers each one on the image, and lets you correct the last few by tapping — so the final number is yours, with the picture to back it up.
- Runs entirely in your browser
- Works offline
- No upload
- No sign-up
- No watermark
Loading the tool…
How it works
- Take the photo straight on — the round ends facing the camera, even light, the whole stack in frame.
- Drop it here. The fast mode estimates the object size and counts by itself; for anything not round, switch to the AI mode and drag a box around 1–3 example objects.
- Tap the image to add anything missed or remove a false hit, then download the numbered image.
Why nothing is uploaded
Every operation on this page is done by code running inside your browser tab, using the same engine that renders web pages. The file is read from disk into your tab’s memory, transformed there, and written back out as a download. It is never sent anywhere — not to us, not to a third party.
Verify it yourself
- Open your browser’s developer tools (F12) and select the Network tab.
- Load your file and run the tool.
- The only requests you will see fetch the tool’s own code — and, for a few heavy tools, their open-source engine from a public CDN — plus one small page-view ping to loreatec.jp (page address and title, nothing more). None of them carry your file.
Frequently asked questions
What kind of photo works?
Similar objects of similar size, seen end-on: pipe and rebar bundles from behind, logs on a truck, bottles or cans from above, pills on a tray, a jar of coins spread out. Shoot perpendicular to the ends and keep the light even. What breaks it: strong angles (circles become ellipses of different sizes), mixed object sizes in one shot, and objects half-hidden behind others.
How accurate is it?
On a clean, straight-on photo the automatic count typically lands within a few per cent — and the last few per cent are exactly why the tap correction exists. Every detection is drawn and numbered on the image, so you can audit the result instead of trusting it. Treat the downloaded image as the tally sheet.
Why does the diameter setting matter so much?
The detector looks for circles close to that size, and that single constraint is what keeps a wall of seven hundred pipe ends from turning into fourteen hundred half-guesses. It is estimated automatically from the photo; if the result looks wrong, matching the slider to one object is the first thing to fix.
Does it use AI? Is the photo uploaded?
Nothing is uploaded in either mode — the photo never leaves the device, which matters when it shows a customer’s stock. The fast mode is classical computer vision: a well-known open-source image-processing library, fetched once from a public CDN (about 8 MB, then cached), looks for circles of the size you set. The AI mode additionally downloads a 35 MB counting model once, and it too runs entirely inside this tab — flight mode included.
Can it count mixed or irregular things — boxes, birds, people in a crowd?
Yes — that is what the AI mode is for. Draw a box around one to three examples and a neural model trained on counting estimates a density map of everything that looks like them, then sums it. Two honest caveats: for large piles it is an estimate rather than an exact tally (the heat map shows where the count comes from), and it counts one kind of thing at a time — the kind you boxed. The model is MIT-licensed and named on the licences page.