From images to training data,on your computer.

Annotation is the slow step between your images and a trained model. The Labelme desktop app drafts each shape with AI, and you correct it.

  • SAM2, SAM3, and YOLO-World draft from a click, box, or word.
  • Your images and annotations stay on your computer.
  • It installs without Python and writes plain JSON files.
Try Labelme free
See plans

Available for macOS, Windows, and LinuxTry it free for 3 days without a card. After that you pay once, from $49, and can get a full refund within 14 days.

The Labelme desktop app annotating a synthetic Urban Plaza dataset: a selected dog polygon and the label dialog naming it, shown enlarged in a loupe, with drawing tools, label lists, and dataset files around the canvas.

Labelme annotates any kind of image.

  • AgricultureApples as rectangles
  • MedicalLungs as polygons
  • ElectronicsChip as rectangle
  • PlantsLeaf as polygon
  • IndustrialBearing as circle
  • MicroscopyCells as circles
  • TrafficSign as rectangle
  • ConstructionExcavator as rectangle

Labelme turns your images into annotations you can train on.

Your images
Labelme
The Labelme desktop app annotating a synthetic Urban Plaza dataset: a selected dog polygon and the label dialog naming it, shown enlarged in a loupe, with drawing tools, label lists, and dataset files around the canvas.
AnnotateBy handAI draftReview
Labelme JSON
or convert
Pro toolkitYOLOYOLO-OBBVOC
PyTorchTensorFlow
Train
Your AI model
Works with your training stack
  • PyTorch
  • TensorFlow
  • Ultralytics YOLO
  • PaddlePaddle
  • OpenMMLab
  • Roboflow

Point, box, or words. AI drafts the annotation.

You correct shapes instead of drawing them. Every result stays editable, point by point.

A Labelme parking-lot view with SAM3 oriented boxes around every visible car.

Checked, converted, ready to train.

You check the dataset in the app, then give your training code a format it reads.

Reviewing the annotated orchard in Labelme, zoomed into one selected tree-crown polygon, with a QA flag panel and the full annotation checklist.
ReviewMark an image as checked with a flag, and go through every shape in the annotation list.
Exporting the annotated orchard to Pascal VOC and YOLO with the Labelme toolkit: the terminal commands above the class segmentation masks they produced, tree crowns and truck each in their class color.
ConvertLabelme JSON works as it is. Conversion to YOLO, YOLO-OBB, or Pascal VOC needs Pro and runs in a terminal with Python.

Your images stay on your computer.

Everything above runs on one computer. The app uploads nothing, which is what medical, industrial, and client images need.

  • DataImages, annotations, and exports stay in your folders. The app sends no telemetry and processes nothing in the cloud.
  • Local AIThe models run on your computer, without an API key or a third-party AI account.
  • DownloadsYou download the app with a free labelme.io account, and each model on first use. After that, you can work offline.
Air-gapped or behind a firewall? See the FAQ
Offline
Labelme annotating a synthetic medical image on the desktop.

Open a folder and start annotating.

Everything in one window.

There is nothing to configure before the first image.

The Labelme window: tool column, canvas, label list, and file list.

7shape tools.

Pick a tool with one click, then draw.

  • Polygon
  • Rectangle
  • Oriented rectangle
  • Circle
  • Point
  • Line
  • Line strip

One key to the next image.

  • DNext
  • ABack
  • THide shapes

Saves as you work.

There is no save step.

  • On every edit
  • On image switch

21languages.

English简体中文繁體中文日本語한국어DeutschFrançaisEspañolPortuguêsItalianoNederlandsPolskiРусскийУкраїнськаTürkçeΕλληνικάMagyarBahasa IndonesiaTiếng Việtไทยفارسی

Flags and notes.

Mark an annotation occluded or truncated, or add a note.

AI when you want it.

Or draw every shape by hand.

A Labelme parking-lot view with SAM3 oriented boxes around every visible car.

Plain JSON files.

Labelme writes one beside each image, and any script can read it.

{"imagePath":
   "urban_plaza_0001.png",
 "shapes": [{"label": "dog",
   "shape_type": "polygon",
   "points": [[832.4, 861.0], …]}]}

Already use open-source Labelme? Skip the setup.

The open-source package has the same tools and models, and you set it up yourself. The Labelme app does that setup for you.

Open-source packageLabelme app
InstallSet up Python, then run pip install labelmeOne installer for macOS (Apple Silicon), Windows, or Linux. Double-click it to open the app. It needs no Python.
Model downloadsWeights download from GitHub and Hugging Face, which filtered networks can blockWeights download from the Labelme mirror, so a filtered network needs to allow one host
Updates & supportYou upgrade with pip and resolve package conflicts yourselfYou get a year of tested builds, and your version keeps working after that. Email support is included.
CostThere is no license fee. Setup and upkeep cost your time.One payment from $49

"I used the free version before, and saved a lot of hassle, now, this wouldn't be much to invest."

Yoon Young JoonVerified purchase
Try Labelme freeSee plans

Rated 4.9 from 146 reviews.

Every review is from a buyer, quoted as written.Read them all on Gumroad

5 out of 5 stars

"The best thing about it is that it is easy to annotate, even for the average person who is not a programmer. Other apps require writing code or complicated steps, but this has none of that."

WakuiMoritoshi
5 out of 5 stars

"Very awesome tool, helped me create lots of datasets by myself in a very efficient and quick way! Easy to use and performs outstandingly, very happy"

John Smith
5 out of 5 stars

"The best standalone annotation tool."

Royi Avital
5 out of 5 stars

"Super useful for image annotation in Windows."

Gumroad review

From the person who wrote Labelme.

A crowd watching a red two-armed robot pick items at the Amazon Picking Challenge.
The Amazon Picking Challenge, where Labelme began.

I wrote the first version of Labelme in 2016, as a student in a robotics lab at the University of Tokyo. Our team was entering the Amazon Picking Challenge, and I needed annotated images to train the robot's vision. It has been open source ever since, and I still maintain it.

The app on this page is that same Labelme, built and tested so you can skip the setup. Buying it pays for that work and keeps the open-source project maintained. If it does not work for you, email support@labelme.io within 14 days and I will refund you in full.

Kentaro WadaAuthor of Labelmekentaro@labelme.ioPhD from Imperial College London. Formerly computer vision lead at Mujin. Alumnus of the JSK Robotics Lab, University of Tokyo.

You pay once and keep the app.

Try it free for 3 days without a card. After that you pay once, from $49, and can get a full refund within 14 days.

  • Starter

    Annotate images by hand or with AI.

    $49USDone-time
    Get Starter

    Includes 1 year of updates. Your version keeps working.

    By purchasing, you agree to the EULA.

    • The Labelme app for macOS, Windows, and Linuxv7.8.0
    • AI annotation from a click (SAM2)
    • AI annotation from text (SAM3, YOLO-World)
    • Interface in 21 languages, including Chinese and Japanese

  • Pro

    Most popular

    Convert and batch-process whole datasets.

    $79USDone-time
    Get Pro

    Includes 1 year of updates. Your version keeps working.

    By purchasing, you agree to the EULA.

    • Everything in Starter, plus:
    • Dataset toolkit: 10+ command-line tools to batch-process, review, and convert
    • Export to YOLO, YOLO-OBB, and Pascal VOC
    • Advanced guides for dataset creation and model training
    • Priority email support, answered within 48 hours

    Toolkit tools run in the terminal and require Python.

    Explore the dataset tools

  • Pro (Lifetime)

    Pro, plus every future version.

    $249USDone-time
    Get Lifetime

    Includes every future update.

    By purchasing, you agree to the EULA.

    • Everything in Pro
    • Every future version, major releases included
    About the price of 3 years of Pro

Refund by email within 14 days, no questions asked.

Instant access after purchasePay by card, PayPal, Apple Pay, or Google Pay

Secure payment via PaddleVAT may apply.

Team license

We arrange team and enterprise licenses by email. Every seat gets Pro, on one invoice at volume pricing.

We'll get back to you within 1 business day

Contact us

Questions and answers.

Check anything before you pay. Start with the trial, the price, the refund, and where your images go.

Before you buy

What's the difference between open source, Starter, and Pro?

The open-source installation gives you the Python application, with the same annotation tools and AI models. Starter is the tested desktop build of it, so there is no Python setup. Model weights download through our own mirror, which works behind restrictive firewalls. Pro adds the dataset toolkit, conversion to YOLO, YOLO-OBB, and Pascal VOC, advanced guides, and priority email support. Compare the plans.

Can I try Labelme before buying?

Yes. The Labelme app runs free for three days with no credit card. To get it, download it here with a free labelme.io account. A new account does not restart the trial, and the Pro dataset toolkit is not part of it.

Is Labelme a subscription?

No. Every plan is a one-time purchase with no recurring fee. Starter and Pro include one year of updates, and the version you have keeps working after that. Pro Lifetime includes every future update. Labelme runs on your computer with no cloud behind it, so there is nothing to subscribe to. Read why offline-first annotation matters for the reasoning.

Do my images ever leave my computer?

No. Images, annotations, and exports stay on your computer. The app uploads nothing, processes nothing in the cloud, and sends no telemetry. It downloads model weights on first use.

What is the refund policy?

Fourteen days, no questions asked. Email support@labelme.io within fourteen days of payment and we refund in full through Paddle to your original payment method. A refund ends your access.

Choosing Labelme

Is Labelme open source?

Yes. The annotation application is GPL-3.0, maintained on GitHub, and installable with pip. We build the Labelme app sold here from that source. From 7.8.0 it ships under the EULA rather than the GPL, with the tested binary and the model mirror. Pro also adds the dataset toolkit, which is closed source.

When is Labelme not the right tool?

If you are comfortable managing Python and your network reaches GitHub and Hugging Face, the open-source package gives you the same annotation tools and AI models for no license fee. Labelme is also a desktop app with no cloud behind it, so there is no shared online workspace.

What is Labelme?

Labelme is an image annotation tool for creating training datasets. The open-source route is a Python application. The Labelme app packages it as a standalone desktop app for macOS, Windows, and Linux, with AI models that download once and then run on your computer.

Is this the same as MIT CSAIL LabelMe?

No. Labelme by Kentaro Wada is distinct from the MIT CSAIL LabelMe project. This site is for the open-source application and its paid offline desktop build.

Privacy and AI

Does Labelme work without an internet connection?

Yes, after a one-time download per model. The app fetches each AI model on first use, and after that annotation and AI assistance run on your computer with no connection. If your network filters outbound traffic, allow osam-models.labelme.io on port 443. The Labelme app fetches models from there first.

Can I run Labelme fully air-gapped?

Yes, with one step. Download the models on a connected computer first, or ask support@labelme.io for the offline pack.

Do I need an API key or cloud account?

No API key and no third-party AI account. You need a free labelme.io account to download the app and its updates. The app itself never signs in. It asks once for your license key and checks it offline.

Which AI models are supported?

Labelme supports the SAM family (SAM, SAM2, SAM3, and EfficientSAM) for assisted segmentation and YOLO-World for text-prompt object detection. The models run on your computer rather than through a hosted API. In Labelme, SAM3 takes box and text prompts, so use SAM2 or EfficientSAM for point prompts. YOLO-World returns bounding boxes.

Annotation workflow

Which operating systems are supported?

The desktop app runs on macOS (Apple Silicon), Windows (64-bit), and Linux (64-bit). Intel Macs are supported up to v7.0.4.

Which export formats are available?

Pro includes automated conversion for YOLO, YOLO-OBB, and Pascal VOC, plus import back from YOLO and YOLO-OBB. The toolkit also produces masks, visualizations, and dataset statistics from the command line.

What does the Pro dataset toolkit add?

The toolkit has tools for batch processing, visual review, resizing images together with their annotations, label renaming, dataset statistics, export, and CLI automation.

What annotation types can I create?

Labelme supports polygons, rectangles, oriented rectangles, circles, lines, points, masks, flags, and notes. AI assistance proposes masks and polygons, and you edit the final shape.

Plans, purchase, and licensing

Can I upgrade from Starter to Pro later?

Not at a discount. Pro is a separate one-time purchase at its regular price, with no credit for Starter. If you expect to need the dataset toolkit, starting with Pro costs less than buying both.

What happens when my year of updates ends?

Nothing stops working. Your license key keeps unlocking every release from your year of updates, so the version you have keeps running indefinitely. You only pay again if you want newer releases, or choose Pro Lifetime, whose key unlocks all of them.

Can I use my license on more than one computer?

Yes. A license covers one person, not one device. Your key works on all your own computers, and the app checks it offline, so there is no activation server.

What happens after I purchase?

You get access as soon as payment clears. Pay with the same email you want to sign in with. If you do not have an account yet, the purchase creates one and sends a link to set a password. Then sign in to download the app.

Can I use Labelme commercially?

Yes, for your own or your organization's internal annotation and dataset work. From 7.8.0, we license the desktop app under the End-User License Agreement. The source code and pip package stay GPL-3.0. We license the desktop app, the Pro toolkit, and other paid features to one named user. You may not redistribute, resell, or sublicense them.

Can I purchase Labelme for a team or organization?

Yes. We arrange team licenses, purchase orders, and invoices by email. Contact support@labelme.io and we will set it up. We support academic, institutional, and reseller purchases. Each seat belongs to one named person, so shared accounts are not permitted.

Can I receive an invoice or claim tax exemption?

Paddle sends a receipt and invoice after purchase. Enter your VAT number at checkout to apply EU reverse charge, because it cannot be applied afterwards. Buyers elsewhere can send exemption documents to assist@paddle.com.

What licenses apply to the app and the AI models?

From 7.8.0, we license the desktop app under the End-User License Agreement. The source code and pip package stay GPL-3.0. The AI models carry their own licenses: SAM, SAM2, and EfficientSAM are Apache-2.0, SAM3 is under Meta's SAM License, and YOLO-World is GPL-3.0. Contact support@labelme.io if your organization needs the license texts for review.

Annotate your next dataset on your computer.

AI drafts each shape, your images stay on your computer, and one installer replaces the Python setup.

Try Labelme free
See plans

Available for macOS, Windows, and LinuxTry it free for 3 days without a card. After that you pay once, from $49, and can get a full refund within 14 days.