Pixano

Overview

Pixano is an open-source computer-vision dataset tool for teams that need to explore, search and annotate image or video collections. It is best suited to developers who can work with Python and want annotation components and APIs they can build into their own workflow. Its breadth is appealing, but active development and possible API changes make it a less settled foundation for integrations that must stay stable.

Key features

Annotation and model assistance

Pixano covers bounding boxes, editable polygons, pixel masks, keypoints, cuboids, classification and tracking, with customizable labels and propagation of video annotations over time. That range gives teams a route from common 2D labeling to temporal and 3D tasks within one tool. Cuboids can be matched to point clouds using geometric transformations, although point-cloud support is described as planned for the broader multi-view dataset experience.

Model-assisted labeling includes smart segmentation using models such as SAM; semantic search using models such as CLIP can help locate relevant material in a dataset. These features are most useful for AI developers who already work with such models, rather than teams seeking a turnkey labeling service.

Datasets, APIs and customization

Pixano supports multi-view datasets containing text, images and video, and imports and exports formats such as COCO. Annotation exports include COCO and Pixano's own format; JPEG and PNG are supported image formats. Lance is used for storage and dataset navigation, while REST and Python APIs offer ways to interact with the application and datasets. Reusable annotation elements are Web Components, so developers can assemble a custom app instead of treating the interface as fixed.

Pixano Inference adds a Ray Serve-based server for deployed models, with a Python client and REST API. This developer-oriented integration is a strength for teams connecting labeling and inference, but also increases the importance of accounting for API changes while the project is under active development.

Pricing

Pixano is free, open-source software under the CeCILL-C license. There is a free plan and no free trial; no paid plan or seat-based tier is part of its pricing model. The absence of a subscription makes it suitable for teams that can deploy and operate the software themselves, but free access should not be mistaken for a hosted service or a promise of support. The README points users to its Getting Started and contributing guides for usage and contribution information.

Platforms

Pixano is available for web use and can be deployed on Linux, macOS or Windows, with self-hosted and API options. The project documents installation through pip in a Python virtual environment or through official Docker releases. Its recommended Python range is version 3.10 or later and earlier than 3.14. That deployment flexibility suits technical teams, while the documented Python requirement and self-hosting path may be a barrier for users looking for a managed, ready-to-use service.

Who it's for

Pixano is aimed at AI developers and applications in areas such as manufacturing, security, robotics and transportation. It makes sense for teams handling computer-vision datasets that want to combine annotation, model-assisted labeling, search and custom integrations. It is less compelling for organizations that need a stable, managed platform with established security assurances: no security certification or compliance standard is stated, and the project warns that its API may change.

Pros and cons

• Pros: Broad annotation coverage, from polygons and pixel masks to tracking and cuboids, serves varied computer-vision workflows.
• Pros: COCO import and export, plus Pixano-format exports, gives teams practical routes to move annotations between workflows.
• Pros: Python and REST APIs, reusable Web Components and Pixano Inference support developer-led customization and model integration.
• Pros: Free, open-source licensing avoids subscription charges and allows self-hosted deployment.
• Cons: Active development and potential API changes create maintenance risk for teams building long-lived integrations.
• Cons: Installation requires a compatible Python environment or Docker deployment, making it less suited to nontechnical users seeking a managed service.
• Cons: No stated security certification or compliance standard may rule it out for teams with formal assurance requirements.

Alternatives

For a broader comparison, browse Image Annotation Software and AI Image Annotation Tools.

• Label Studio is another option with a free Community Edition and a free trial; consider it if you want to compare a freemium alternative across web, desktop, API and self-hosted platforms.
• Roboflow offers a free tier with 10 credits per month, described as enough for about 30 model training runs or 80,000 inferences, as well as a Core plan at $39 a month. Choose it if those stated allowances or its Android and iOS platform support fit your needs better.
• CVAT has a free Community plan for personal use and small teams under the MIT license, and a separate free online plan with limits of one member, one project, three tasks and 1 GB. It is worth comparing if those defined free options better match your team size and workload.
• Ultralytics Platform has a free plan with 100 GB storage, 10 GB per upload, 100 models, three concurrent cloud training jobs and three cloud deployments. Consider it if those cloud capacities are a better match than Pixano's self-hosted, developer-oriented approach.
• BasicAI offers private-cloud deployment starting from $6,600 a year, with customizable seats, storage, model calls and on-premise deployment options. It is an option to compare if those deployment terms suit your requirements.
• Labelbox is another freemium option with a free plan.
• Labelme is another freemium option for Windows, macOS and Linux.
• MakeSense.ai is another free option for web, Windows, macOS and Linux.

Verdict

Choose Pixano if your team needs a free, open-source toolkit for computer-vision dataset annotation and exploration, and has the technical capacity to deploy it and maintain integrations. Its strongest case is the combination of broad annotation tools, model-assisted features and APIs that can support custom workflows. Look elsewhere if you need a managed service, formal security assurances, or an integration surface that is less exposed to change.