The short answer
Why use Count Objects?
Automatic detection is a useful first pass, but it should not make the final decision. Editable boxes let you correct misses before an image is blurred, counted, or exported.
Count Objects focuses on detecting common objects and reviewing the count before export. A self-hosted model runs in the browser, then returns confidence-scored regions. You can adjust the threshold and add a manual box when the model needs help.
The selected file stays in this browser tab and is released when you leave the page. Detection varies with angle, lighting, size, and the objects represented in the model. Review every box before trusting the export.
What Count Objects does
Open the source photo
Drop, choose, or paste a supported file into Count Objects. Detecting common objects and reviewing the count before export. The source stays on your device; we work on a copy.
Review the detected boxes
Count Objects is for detecting common objects and reviewing the count before export. The preview is the place to decide if the setting went too far. The original file is not overwritten.
Useful controls
Main controls: Confidence threshold plus editable or numeric correction boxes. Working limit: 50 MB and 50 megapixels per image; model loading and inference need more memory than a basic filter. Watch the preview at full size before you save. A current desktop browser usually has more memory than a phone.
Download the reviewed copy
Export is an annotated or blurred image and on-screen detection details. The download is a new file. Keep the original until you have opened the copy once.
Where Count Objects fits
These are ordinary jobs where the focused controls save time. They also show where a full editor or a manual review may still be the better choice.
Privacy checks
Cover a face or plate before posting a photo publicly.
Inventory drafts
Get a quick object count, then correct boxes that overlap or miss items.
Manual review
Use Count Objects as a starting point rather than treating a model result as final.
Start to finish
How to use Count Objects
- 01Choose, drop, or paste a supported file.
- 02Adjust the count objects controls and check the preview at full size.
- 03Download the finished copy and open it once before using it elsewhere.
What to expect from Count Objects
This table covers the parts that matter before you begin: what goes in, which controls do the work, what comes out, and where the browser draws a hard line.
| Check | What it means here |
|---|---|
| Supported input | JPEG, PNG, WebP, GIF, BMP, SVG and browser-supported AVIF images |
| Main controls | Confidence threshold plus editable or numeric correction boxes |
| Output | An annotated or blurred image and on-screen detection details |
| Working limit | 50 MB and 50 megapixels per image; model loading and inference need more memory than a basic filter |
| Privacy | Files and entered values stay in this tab. Filenames, pixels, metadata, OCR text, QR contents, locations, and passwords are not sent to analytics. |
Before you export
- Lower confidence only when you are prepared to remove false boxes.
- Add a manual box for a missed region.
- Review small and partly hidden objects carefully.
- Do not use automated detection as the only privacy check.
Frequently asked questions
If you cannot find what you are looking for, write us a message and we will get back to you.
What should I check before saving with Count Objects?
Does my file get uploaded?
Can I use Count Objects on a phone?
Will Count Objects overwrite my original file?
Why might the Count Objects result look different in another app?
Is Count Objects free to use?
How does Count Objects work?
What can I put into Count Objects?
How do I save the result from Count Objects?
Why did Count Objects fail or look wrong?
Will Count Objects keep a copy of my work?
Related tools
More in Measurement.
Continue the workflow
These tools handle common next steps after count objects.