About
BIIGLE is a next-generation, open-source web platform for annotating and exploring large collections of images and videos. Originally developed for marine environmental monitoring and scientific research, it has grown into a general-purpose annotation system trusted by thousands of scientists and researchers worldwide, with over 40 million annotations and 6 million images and videos processed. The platform supports a wide variety of annotation types including points, rectangles, circles, lines, and polygons, as well as whole-frame and timeline-based annotations for videos. Its efficient workflow is complemented by AI-assisted features: MAIA helps discover and annotate objects of interest automatically, while Magic SAM leverages the Segment Anything Model for rapid object segmentation — dramatically reducing manual annotation time. BIIGLE's Largo label review grid is purpose-built for rapid quality control and exploration of thousands of annotations, optimizing human visual perception for fast review. Users can invite collaborators, create hierarchical label classification schemes, version and share label trees, and import species labels directly from the World Register of Marine Species. For large-format scientific imagery such as gigapixel mosaics, microscopy images, and orthophotos, BIIGLE renders seamlessly in the browser. A comprehensive REST API enables integration with external machine learning pipelines and custom analysis workflows. The source code is freely available under the GPL-3.0 license, and the public instance at biigle.de is hosted free of charge via the de.NBI Cloud.
Key Features
- AI-Assisted Annotation: MAIA and Magic SAM (Segment Anything Model integration) help automatically detect, segment, and annotate objects of interest, dramatically reducing manual effort.
- Multi-Format Annotation Support: Supports points, rectangles, circles, lines, and polygons for images, plus whole-frame and timeline-based tracking annotations for videos.
- Largo Label Review Grid: A visual grid interface optimized for rapid review and quality control of thousands of annotations, designed to maximize human visual perception efficiency.
- Team Collaboration & Label Management: Invite colleagues, assign permission levels, create hierarchical label trees, and version/share classification schemes including species labels from global registries.
- REST API & Pipeline Integration: A comprehensive API enables programmatic access to annotations and data, supporting integration with machine learning pipelines and custom analysis workflows.
Use Cases
- Marine biologists annotating seafloor survey images and videos for species identification and environmental monitoring.
- Research teams performing large-scale quality control review of annotation datasets using the Largo grid interface.
- Data scientists integrating BIIGLE via API into machine learning pipelines to manage training data labeling and inspection.
- Universities and research institutions self-hosting BIIGLE to manage proprietary image datasets with collaborative annotation workflows.
- Ecologists and environmental scientists labeling drone orthophotos or microscopy gigapixel mosaics for habitat and species analysis.
Pros
- Completely Free and Open Source: BIIGLE is free to use on the public instance and fully open source under GPL-3.0, making it accessible to researchers and institutions with any budget.
- Powerful AI Tools Built In: MAIA and Magic SAM reduce annotation time significantly by automating detection and segmentation tasks directly within the platform.
- Scalable for Massive Datasets: Handles millions of images and videos including gigapixel mosaics, with browser-native rendering that stays fast regardless of image size.
- Self-Hosting Option: Organizations can deploy their own BIIGLE instance for full data sovereignty and customized environments.
Cons
- Steep Learning Curve for Advanced Features: Setting up MAIA pipelines or self-hosting a custom instance requires technical knowledge and familiarity with the documentation.
- Primarily Research-Oriented UI: The interface is optimized for scientific workflows and may feel less intuitive for non-research commercial annotation use cases.
- Limited Native Integrations: Unlike some commercial annotation platforms, BIIGLE does not offer out-of-the-box integrations with popular third-party project management or data pipeline tools.
Frequently Asked Questions
Is BIIGLE really free to use?
Yes. The public BIIGLE instance at biigle.de is free of charge, funded by the de.NBI Cloud. The source code is also open source under the GPL-3.0 license, so you can self-host your own instance.
What types of annotations does BIIGLE support?
BIIGLE supports points, rectangles, circles, lines, and polygons for images. For videos, it additionally supports whole-frame annotations and timeline-based annotations that track objects across multiple frames.
How does the AI-assisted annotation work?
BIIGLE includes MAIA, which helps identify and annotate objects of interest using machine learning, and Magic SAM, which uses Meta's Segment Anything Model for fast object segmentation. You can also upload results from offline ML workflows for inspection and quality control.
Can multiple users collaborate on the same project?
Yes. BIIGLE supports team collaboration with different permission levels. Teams can create annotation sessions, share and version label trees, and even import species labels from the World Register of Marine Species.
Does BIIGLE provide API access?
Yes. BIIGLE offers a comprehensive REST API for programmatic access to your data and annotations, enabling integration with external machine learning pipelines and custom analysis workflows.
