Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

LightlyStudio

lightly.ai

  • Free plan
  • Open source
  • Windows
  • macOS
  • Linux
  • Web
  • API
  • Self-hosted
8.5 Club score Visit lightly.ai

What LightlyStudio is

LightlyStudio brings computer-vision dataset curation, annotation, quality checks, evaluation, and management together. Its curation tools compute embeddings and let users filter by metadata, search for similar text or images, remove duplicates, find outliers, and balance classes. Teams can annotate images and video with model-assisted labeling plugins, then review model results using a confusion matrix and per-sample metrics. It reads media from Amazon S3, Google Cloud Storage, and Azure Blob Storage, and works with formats including COCO, YOLO, Pascal VOC, and YouTube-VIS. Exports can include a full dataset or a filtered subset in those formats or CSV. A Python API can index and query datasets, select samples, add annotations and tags, run sampling, and launch the GUI. The open-source edition is free under Apache License 2.0. Enterprise offers hosted or on-premise deployment; its on-premise option can run offline and air-gapped. The local app supports Windows, Linux, and macOS on x86 and ARM, with Python 3.9 or later.

Who it suits

It suits computer-vision teams that need to curate and annotate image or video datasets and evaluate models. Enterprise collaboration, permissions, and managed cloud credentials are available for teams that need them.

Good to know, and what to watch

  • Free open-source version under Apache License 2.0
  • Searches for similar text and images
  • Annotates images and video
  • Exports datasets or filtered subsets
  • Local app supports Windows, Linux, and macOS
  • Open Source plan has no customer support
  • Open Source plan has no team collaboration
  • Local application requires Python 3.9 or later

Lightly-Hosted

—

On-Premise

—

Open Source

Free

Free plan

Getting started

  1. Visit https://lightly.ai/lightly-studio.
  2. Choose the free Open Source plan or review the Lightly-hosted and on-premise Enterprise options.
  3. For local use, run the application on Windows, Linux, or macOS with Python 3.9 or later.
  4. Connect image or video storage from Amazon S3, Google Cloud Storage, or Azure Blob Storage.
  5. Use the Python API to index datasets or start the GUI.

Limits to check

The Open Source plan has no customer support or team collaboration. The Lightly-hosted option includes a 14-day free trial; the on-premise offering includes dedicated engineering support.

The Club's review

LightlyStudio is a computer-vision workspace for finding, labeling, and assessing image and video data. It suits practitioners who need dataset curation alongside annotation, particularly those comfortable with open-source software or evaluating an Enterprise deployment. The breadth is a clear advantage; shared work and customer support require more than the free plan.

Overview

LightlyStudio combines data curation, annotation, quality assurance, model evaluation, and dataset management in one app. That makes it a stronger fit for computer-vision workflows that need to shape and inspect datasets, not just attach labels. The Apache License 2.0 version costs nothing, but has no customer support or team collaboration, so teams needing either should consider Enterprise.

Key features

Curation and dataset management

Automatic embeddings power metadata filters, text and image similarity search, deduplication, outlier detection, and class balancing. In practice, these tools help users find similar or unusual samples and address class imbalances before allocating labeling effort. Lightly reports handling embeddings for more than two million images on a MacBook M1 with 16 GB of memory. That signals a substantial reported workload, though it is not a performance guarantee for other hardware or datasets.

Annotation, evaluation, and exchange

LightlyStudio supports image and video annotation, model-assisted labeling plugins, review workflows, and model evaluation with a confusion matrix and per-sample metrics. This combination is useful when labeling and checking model behavior belong in the same process. It reads media from Amazon S3, Google Cloud Storage, and Azure Blob Storage, and supports COCO, YOLO, Pascal VOC, and YouTube-VIS annotation formats. Users can export complete datasets or filtered subsets in those formats or CSV, giving teams options to move data into other workflows.

Python API and privacy

The Python API can index datasets, query and slice samples, add annotations and tags, run sampling, and launch the GUI. It provides a programmatic route for teams that want to incorporate dataset operations into their own tooling. In the open-source version, images and videos stay in the user's storage; anonymous usage analytics can be disabled with an environment variable. Lightly says it is ISO 27001 certified and GDPR compliant, and that raw media is never stored on its servers.

Pricing

Open Source: free. It includes all open-source features under the Apache License 2.0, but no customer support or team collaboration. This is the clearest choice for individual users and teams able to manage their own setup and work without shared collaboration features.

Lightly-Hosted: custom pricing, with a 14 days free trial. Lightly hosts the service and the plan includes dataset management and team collaboration. It suits teams that want shared work without deploying the service on their own infrastructure.

On-Premise: custom pricing. The deployment runs on customer infrastructure and includes dataset management, team collaboration, and dedicated engineering support. Its fully offline, air-gapped option is relevant to organizations that need that deployment model; the trade-off is a custom-priced Enterprise plan rather than the free open-source option.

Platforms

LightlyStudio supports API, Linux, macOS, self-hosted, web, and Windows. The local application runs on Windows, Linux, and macOS on x86 and ARM processors, with Python 3.9 or later. The mix of local, hosted, and on-premise deployment gives teams flexibility, while Enterprise is the route to Lightly-hosted or on-premise service.

Who it's for

LightlyStudio is a good match for computer-vision practitioners who need to curate and evaluate datasets as well as annotate images or video. Its Python API, cloud-storage connections, and multiple deployment options also make it relevant to teams integrating dataset work into a broader pipeline. It is a weaker fit for users who need customer support or built-in collaboration but cannot take on a custom-priced Enterprise plan.

Pros and cons

  • Pros: Curation, annotation, review, and evaluation sit in one application, reducing the need to split those stages across separate tools.
  • Pros: The free Apache-licensed version, local app, and on-premise option offer choices for teams with different deployment needs.
  • Pros: Python API operations and support for major cloud storage and annotation formats help connect dataset work to existing processes.
  • Cons: Open Source has no customer support or team collaboration, limiting its fit for teams that need supported shared work.
  • Cons: The hosted and on-premise plans use custom pricing, so their cost cannot be compared from a fixed published rate.

Alternatives

For a broader set of data labeling software, compare tools by the task and deployment model you need. Label Studio is another freemium option with a free plan and API, desktop, web, and self-hosted platform support. Datasaur may fit a smaller labeling setup: its free tier is limited to one user, 5,000 labels per year, 100 MB of storage, and a personal workspace, with a Growth trial of up to 14 days. Segments.ai has a free academic license for non-commercial academic research, subject to a license request. Doccano is a free open-source annotation tool installable with pip, Docker, or Docker Compose. Supervisely offers a free Community plan with 5 GB storage, 10,000 files, and two members, with unlimited projects and annotations. Potato is a self-hosted free option with all features included and no usage limits. MakeSense.ai is a free browser-based photo labeling tool.

Verdict

Choose LightlyStudio if your computer-vision workflow needs dataset curation, annotation, and evaluation together, and the free self-managed version or an Enterprise deployment fits your needs. Its strongest reason to choose it is that breadth, backed by a Python API and flexible deployment. Look elsewhere if you need supported team collaboration without moving to custom-priced Enterprise.

The verdict

LightlyStudio covers dataset preparation, labeling, and evaluation in one app, with a Python API and multiple deployment options. Check the plan details if your team needs support or collaboration, since those are not included in Open Source.

Questions people ask

Is LightlyStudio free?

The Open Source plan is free and is distributed under the Apache License 2.0. A 14-day free trial is listed for the Lightly-hosted option.

What platforms does it support?

LightlyStudio is listed for API, Linux, macOS, self-hosted, web, and Windows. The local application supports Windows, Linux, and macOS, on x86 and ARM processors, with Python 3.9 or later.

Can it annotate images and video?

Yes. It supports image and video annotation, model-assisted labeling plugins, and evaluation using a confusion matrix and per-sample metrics.

Which storage and annotation formats can it use?

It reads media from Amazon S3, Google Cloud Storage, and Azure Blob Storage. Supported annotation formats include COCO, YOLO, Pascal VOC, and YouTube-VIS; exports can also be CSV.

Is LightlyStudio open source?

Its Open Source version is free and distributed under the Apache License 2.0.

What does Enterprise add?

Enterprise adds multi-user collaboration, role-based permissions, and centrally managed cloud credentials. It is available as a Lightly-hosted service or on-premise, with dedicated engineering support for on-premise.

What the maker publishes

Annotation
The product supports classification, object-detection and segmentation annotations that can be edited in the GUI or through Python. source
Annotation and evaluation
It supports image and video annotation, model-assisted labeling plugins, and model evaluation with a confusion matrix and per-sample metrics. source
API and SDK
The Python API can index datasets, query and slice samples, manipulate annotations and tags, run sampling, export data and start the GUI. source
CLI
The command-line package provides quickstart and gui subcommands. source
Cloud integrations
The software can read media from AWS S3, Google Cloud Storage and Azure Blob Storage through optional cloud-storage dependencies. source
Cloud limitation
Cloud storage currently supports raw media folders and COCO imports; other importers still require local files. source
Collaboration
Enterprise collaboration provides Viewer, Labeler, Editor and Admin roles plus centrally managed cloud credentials. source
Company founding
Lightly says it was founded in 2019 from ETH Zurich research. source
Company headquarters
LinkedIn lists Zurich as headquarters and gives a primary Zurich address at Schützengasse 23, 8006 Switzerland. source
Company size
LinkedIn lists Lightly as a privately held software-development company with 11–50 employees. source
Compliance claim
Lightly's LinkedIn profile states that the company is ISO 27001 certified and GDPR compliant. source
Curation
It automatically computes embeddings and supports metadata filters, text and image similarity search, deduplication, outlier detection, and class balancing. source
Curation
It automatically creates embeddings and supports metadata filtering, text or image similarity search, deduplication, outlier detection, diversity sampling and class balancing. source
Deployment
LightlyStudio Enterprise is offered as a Lightly-hosted service or as an on-premise deployment; the on-premise option can run fully offline and air-gapped. source
Documentation depth
The documentation includes quickstarts, dataset setup, workflows, tutorials, API references, glossary pages and enterprise deployment guidance. source
Export
Users can export full datasets or filtered subsets to COCO, YOLO, Pascal VOC, YouTube-VIS, or CSV. source
Export formats
Exports include CSV, COCO detection and segmentation, YOLO, Pascal VOC, COCO captions and YouTube-VIS video segmentation. source
Founded
2019 source
Funding
Lightly's About page states that the company has raised $10 million. source
Headquarters
Zurich, Switzerland source
Hosted data handling
Hosted Enterprise reads media from cloud storage and stores metadata in PostgreSQL on Lightly infrastructure. source
Independent feature summary
SourceForge describes active-learning and self-supervised algorithms that combine predictions, embeddings and metadata for data selection, with API automation and cloud-bucket connectivity. source
Integrations
It reads images and videos from Amazon S3, Google Cloud Storage, and Azure Blob Storage, and supports common annotation formats including COCO, YOLO, Pascal VOC, and YouTube-VIS. source
Limit: documentation and UI
A G2 reviewer asked for more detail in complex-use-case documentation and a more intuitive, polished interface. source

Checked against the maker's pages on 14 Oct 2026.