V7 Darwin is a commercial annotation and training-data platform focused on high-quality image and video labeling, with particular strength in pixel-precise segmentation. It positions itself as a data engine for teams building and iterating on vision models, aiming to automate the most time-consuming parts of dataset creation.
What it does
Annotators produce bounding boxes, polygons, masks, and keypoints for detection, instance and semantic segmentation, and classification. Its Auto-Annotate feature, built on the Segment Anything Model, generates pixel-precise masks for complex shapes, and video labeling is accelerated through automatic object tracking across frames. Model-in-the-loop capabilities let teams plug in their own models to pre-label new data.
Key features
- SAM-based Auto-Annotate for pixel-precise masks and segmentation
- AI-assisted video auto-tracking across frames
- Support for images, video, and volumetric medical data (DICOM/NIfTI)
- Configurable multi-stage workflows with conditional logic, consensus, and task assignment
- Model-in-the-loop pre-labeling and API access
Deployment and users
Darwin is delivered as a managed cloud platform, with enterprise deployment options, under commercial subscription pricing rather than an open-source license. It is commonly used by computer-vision teams in domains such as healthcare, life sciences, agriculture, and manufacturing that need precise segmentation and structured review pipelines.
ImageVideo