Label Studio From $99/mo · Free plan 8.9 Visit doccano
github.com
What it is
doccano is a free, open-source annotation tool for labeling data for human review and machine-learning work. It supports text classification, sequence labeling and sequence-to-sequence tasks, with examples including sentiment analysis, named-entity recognition and summarisation. A project workflow lets users create projects, import datasets, add collaborators, set annotation guidelines, label data and export results. Collaboration features include multi-language and mobile support, emoji support and a dark theme. REST APIs connect scripts and machine-learning models, and supported modalities include text, image and audio. The project documents installation with pip, Docker and Docker Compose on Linux, Windows or macOS machines running Python 3.8 or newer. Deployment options include AWS, Heroku or Docker elsewhere. Imported datasets can use Amazon S3 or Google Cloud Storage. SQLite 3 is the default database, while PostgreSQL and other database systems are also documented. Celery handles long-running imports and exports, with SQLite3, RabbitMQ and Redis listed as message-broker options. The installation documentation warns that upgrading an SQLite3 installation can result in database loss. The GitHub repository reports no detected SECURITY.md policy and no published security advisories.
Spec sheet
- What it does
- doccano is an open-source tool for creating labeled datasets for machine-learning applications.Read 15 Oct 2026 · github.com
- Purpose
- doccano is an open-source text annotation tool for humans and machine-learning practitioners.Read 30 Sep 2026 · github.com
- API
- doccano provides REST APIs for integrating it with scripts and machine-learning models.Read 30 Sep 2026 · doccano.github.io
- Support
- The documentation directs users who are stuck to the FAQ and says help and feedback can be sent to the author.Read 30 Sep 2026 · doccano.github.io
- Auto labeling
- The developer documentation includes a dedicated auto-labeling application and configuration area.Read 15 Oct 2026 · doccano.github.io
- Background processing
- Celery handles long-running dataset import and export jobs, with RabbitMQ or Redis available as message brokers.Read 15 Oct 2026 · doccano.github.io
- Browser requirements
- The project tests the latest Chrome and expects current Chrome and Safari versions to work.Read 15 Oct 2026 · doccano.github.io
- Cloud deployment
- The project documents one-click deployment options for AWS and Heroku and deployment anywhere by Docker.Read 30 Sep 2026 · github.com
- Cloud storage integrations
- Official documentation describes storing imported datasets in Amazon S3 or Google Cloud Storage.Read 15 Oct 2026 · doccano.github.io
- Collaboration
- Multiple users can collaborate on annotation projects.Read 15 Oct 2026 · github.com
- Dark theme
- The interface includes a dark theme.Read 15 Oct 2026 · github.com
- Database options
- SQLite 3 is the default database, and the documentation also describes PostgreSQL and other database systems.Read 30 Sep 2026 · github.com
- Default port
- The web server uses port 8000 unless another port is specified.Read 15 Oct 2026 · doccano.github.io
- Deployment options
- Installation is documented through pip, Docker, Docker Compose, source checkout and one-click AWS or Heroku deployment.Read 15 Oct 2026 · doccano.github.io
- Documentation depth
- The documentation covers installation, tutorials, cloud storage, OAuth, auto labeling, developer structure, FAQs, roadmap and contribution guidance.Read 15 Oct 2026 · doccano.github.io
- Emoji support
- The feature list states that emoji characters are supported.Read 15 Oct 2026 · github.com
- Installation
- The official project documents installation with pip, Docker, and Docker Compose.Read 30 Sep 2026 · github.com
- Latest release
- Version 1.8.5 was released on January 11, 2026 and included dependency, Docker, assignment and filtering changes.Read 15 Oct 2026 · github.com
- Licence
- The software is distributed under the MIT licence, allowing use, modification, redistribution and commercial sale subject to licence notices.Read 15 Oct 2026 · github.com
- Limit: maintenance cadence
- An independent review highlights the long gap between the July 2023 1.8.4 release and the January 2026 1.8.5 release.Read 15 Oct 2026 · hysenlabs.com
- Limit: operational overhead
- An independent review notes that the pip deployment requires both a web server and a separate task-queue process.Read 15 Oct 2026 · hysenlabs.com
- Limit: pre-labeling fit
- That Reddit user said doccano's built-in pre-labeling did not work for their spaCy NER use case and they uploaded prelabels through the API instead.Read 15 Oct 2026 · reddit.com
- Limit: text-focused scope
- An independent technical review characterizes doccano as text-focused and unsuitable when image, audio or video annotation is required.Read 15 Oct 2026 · hysenlabs.com
- Maker announcement
- TIS announced doccano's open-source release on November 6, 2018 for producing labeled NLP and machine-learning data.Read 15 Oct 2026 · tisi.jp
- Maker background
- TIS describes itself as an IT services company serving more than 3,000 business customers and belonging to a group of about 60 companies with roughly 20,000 people.Read 15 Oct 2026 · tisi.jp
- Mobile support
- The project lists mobile support for its web interface.Read 15 Oct 2026 · github.com
- Multilingual support
- The feature list states that annotation can be performed in multiple languages.Read 15 Oct 2026 · github.com
- OAuth integrations
- OAuth social login is documented for GitHub and Active Directory.Read 15 Oct 2026 · doccano.github.io
- Open issues
- The public issue tracker shows 355 open issues, including reports about Windows installation, Docker Compose startup, exports that run indefinitely and hard-coded secret-key fallback.Read 15 Oct 2026 · github.com
- Package metadata
- PyPI lists Hironsan as the sole maintainer, requires Python 3.8 through before 4.0, and classifies the package as MIT licensed.Read 15 Oct 2026 · pypi.org
- Praise: collaborative workflow
- The same Reddit discussion describes using doccano collaboratively for many kinds of labeling.Read 15 Oct 2026 · reddit.com
- Praise: ease of self-hosting
- A Reddit user reported that self-hosting was very easy and that the team could label collaboratively.Read 15 Oct 2026 · reddit.com
- Python client
- The separate doccano-client project provides a Python client for logging in, listing projects and calling the API.Read 15 Oct 2026 · github.com
- REST API
- A RESTful API is provided for connecting scripts and machine-learning workflows.Read 15 Oct 2026 · doccano.github.io
- Roles
- The backend defines administrator, annotator and approver project roles.Read 15 Oct 2026 · doccano.github.io
- Runtime requirement
- The documentation says doccano can be installed on Linux, Windows, or macOS machines running Python 3.8 or newer.Read 30 Sep 2026 · doccano.github.io
- Security configuration
- Administrators are told to replace the fixed default SECRET_KEY with a unique unpredictable value and can enable file-type checks for imports.Read 15 Oct 2026 · doccano.github.io
- Security status
- The GitHub repository says no SECURITY.md security policy has been detected and no security advisories have been published.Read 30 Sep 2026 · github.com
- Sequence labeling
- The product supports sequence-labeling tasks such as named-entity annotation.Read 15 Oct 2026 · github.com
- Sequence-to-sequence
- The product supports sequence-to-sequence annotation used for tasks such as summarization and translation.Read 15 Oct 2026 · github.com
- Storage integrations
- Supported cloud storage backends for imported datasets are Amazon S3 and Google Cloud Storage.Read 30 Sep 2026 · doccano.github.io
- Supported databases
- SQLite 3 is the default database, while PostgreSQL and other database systems can be configured.Read 15 Oct 2026 · doccano.github.io
- System requirements
- The installation guide requires Python 3.8 or newer on Linux, Windows or macOS.Read 15 Oct 2026 · doccano.github.io
- Task queue integrations
- doccano uses Celery for long-running import and export tasks and documents SQLite3, RabbitMQ, and Redis as message-broker options.Read 30 Sep 2026 · doccano.github.io
- Task types
- It supports text classification, sequence labeling, and sequence-to-sequence tasks.Read 30 Sep 2026 · github.com
- Text classification
- The product supports text-classification annotation projects.Read 15 Oct 2026 · github.com
- Upgrade limitation
- The installation documentation cautions that upgrading a SQLite3 installation can lose its database.Read 30 Sep 2026 · doccano.github.io
- Upgrade warning
- The documentation warns that upgrading an installation using SQLite 3 can lose its database.Read 15 Oct 2026 · doccano.github.io
- Use cases
- The project lists sentiment analysis, named-entity recognition, and text summarization as examples.Read 30 Sep 2026 · github.com
- Workflow
- Users can create projects, import datasets, add users, define annotation guidelines, annotate data, and export labeled datasets.Read 30 Sep 2026 · doccano.github.io
Plans and prices
doccano open-source self-hosted Free text annotation · self-managed infrastructure · pip, Docker or Docker Compose Read 15 Oct 2026 · github.com
MIT-licensed software Free free of charge · use, copy, modify, publish, distribute, sublicense, sell Read 30 Sep 2026 · github.com
Strong points and limits
Strong points
- Supports classification, sequence labeling and sequence-to-sequence tasks.
- Includes collaboration and REST API access.
- Can be installed with pip, Docker or Docker Compose.
- Supports text, image and audio modalities.
- Free and open source under the MIT license.
Limits
- Upgrading a SQLite3 installation can lose its database.
- No SECURITY.md security policy is detected.
- No security advisories have been published.
Who it suits
It suits teams and practitioners who need to create and manage annotation projects for machine-learning datasets, including collaborative labeling and API-based workflows.
doccano in depth
Read the full write-up
Overview
doccano is an open-source annotation tool for teams preparing labeled data for machine-learning work. It suits practitioners who want to manage annotation projects themselves and connect them to scripts or models. Its appeal is a free, adaptable workflow; its operational trade-off is that installation, deployment, and database upkeep are yours to manage.
Key features
Its documented task types—text classification, sequence labeling, and sequence-to-sequence work—cover common ways to structure text labels, including sentiment analysis, named-entity recognition, and summarization. That focus makes doccano a practical fit for text datasets, but the task list does not establish equivalent workflows for every modality the platform supports.
Projects take users from importing a dataset through adding annotators, setting guidelines, labeling, and exporting the result. Collaborative annotation and human review workflows give teams a way to coordinate and check work; custom ontologies and model-assisted labeling provide further support for tailored projects. Multi-language and mobile support, emoji support, and a dark theme round out the interface options.
REST APIs connect doccano to scripts and machine-learning models, which helps teams incorporate labeling into an existing pipeline. Installation is documented through pip, Docker, and Docker Compose; deployment options include Docker anywhere and one-click routes for AWS and Heroku. Imported datasets can use Amazon S3 or Google Cloud Storage. Celery handles long-running imports and exports, with SQLite3, RabbitMQ, and Redis documented as message-broker options.
SQLite 3 is the default database, while PostgreSQL and other database systems are also described. The key caution is concrete: upgrading a SQLite3 installation can lose its database. Teams using SQLite should account for that risk before upgrading; those needing another database have documented alternatives. The repository reports no detected SECURITY.md policy and no published security advisories, so organizations with formal security-policy requirements should weigh that status in their deployment decision.
Pricing
doccano is free, open-source software under the MIT license. The listed MIT-licensed software plan costs nothing, with permission to use, copy, modify, publish, distribute, sublicense, and sell. There is no paid tier or seat-based price to weigh; the trade-off is that self-hosted teams take responsibility for operating the software and its database.
Platforms
doccano is available on web, Linux, macOS, Windows, Android, and iOS, and supports API access and self-hosting. The documentation says it can be installed on Linux, Windows, or macOS machines running Python 3.8 or newer. Pip, Docker, and Docker Compose provide documented installation paths, while AWS and Heroku have one-click deployment options. This range gives technical teams flexibility, though the Python requirement and deployment choices make it a better fit for teams prepared to manage an installation than for buyers seeking a fully managed service.
Who it's for
Choose doccano if you need a free, self-hosted way to coordinate data annotation, define project guidelines, review human labeling, and export datasets, particularly for text-focused machine-learning work. Its APIs, storage integrations, and database options also suit teams fitting annotation into existing systems. Look elsewhere if you need a managed service or cannot accept the SQLite upgrade risk without moving to another documented database.
Pros and cons
Pro: The project workflow spans import, team setup, guidelines, annotation, and export, giving teams a clear route from dataset to labeled output.
Pro: REST APIs, model-assisted labeling, and several deployment and database options support integration and operational choice.
Pro: The MIT license allows broad reuse, including modification and distribution, at no software charge.
Con: SQLite upgrades can lose the database, a material risk for installations that retain the default.
Con: Self-hosting and multiple deployment routes mean the team must take responsibility for installation and operation.
Con: The repository has no detected security policy or published advisories, which may not satisfy organizations that require a published policy.
Alternatives
AI Data Labeling Tools is the broader category for comparing options. Choose Label Studio if you want a freemium alternative with a free plan and trial, across web, API, desktop, and self-hosted platforms. Amazon SageMaker Autopilot is a paid option with a trial and pay-as-you-go pricing without minimum fees or upfront commitments; consider it when that usage-based model fits better than doccano's self-hosted, open-source approach.
Roboflow offers a free tier with 10 monthly credits, enough for about 30 model trainings or 80,000 inferences, and a Core plan at $39 a month. Supervisely has a free Community plan with 5 GB storage, 10,000 files, two members, and unlimited projects and annotations, as well as a Pro plan from 199 EUR per month. These are worth comparing when their stated plan limits suit your team better than running doccano yourself.
BRAT is another free, MIT-licensed, self-hosted option. Prodigy is a paid alternative with a 390.00 USD one-time Personal plan and a free trial. Alibaba Cloud PAI iTAG offers free manual labeling, with OSS storage and data transfer billed separately. CVAT has a free, MIT-licensed Community plan for personal use and small teams, alongside a free online plan capped at one member, one project, three tasks, and 1 GB of storage. Each is an alternative to compare when its stated licensing, pricing, or plan limits better match your needs.
Verdict
doccano is a strong choice for technical teams that want free, self-hosted annotation with collaborative workflows, APIs, and control over deployment. Its main reason to choose is the combination of an end-to-end labeling workflow and an adaptable open-source setup. Look elsewhere if you need managed operations, or plan to keep SQLite and cannot accept the documented upgrade risk.
Verdict
doccano provides project workflows for annotation and export, with multiple deployment and database options documented. SQLite users should heed the upgrade warning, and the repository reports no detected security policy.
Where Specifiction places it
- AI Data Labeling Tools#2 of 228.8
- Api Docs#3 of 198.9
