# V7 Darwin – Data Labeling for Frontier AI Models

> V7 Darwin is a training data platform by V7 Labs for computer vision, medical imaging, and RLHF projects. Teams create high-quality ground truth 10x faster using AI-assisted annotation, multi-stage review workflows, and model-in-the-loop pipelines. SOC 2 Type II and HIPAA compliant. Rated 4.8 on G2.

Darwin serves three main groups: ML engineers and AI teams building computer vision models (automotive, manufacturing, logistics, robotics), medical AI companies annotating DICOM, NIfTI, and whole slide imaging (radiology, pathology, surgery), and AI labs running RLHF or RLAIF pipelines for language model training. For platform and general tooling questions, start with the Platform Overview. For specific data modalities, use Image & Video Annotation or Medical Imaging. For integration and programmatic access, use Developer Resources.

## Platform Overview

- [V7 Darwin Platform](https://www.v7darwin.com/): Training data platform overview — AI-assisted annotation, custom multi-stage review workflows, model-in-the-loop, consensus QA for inter-annotator agreement measurement, and on-demand labeling services; used by leading AI research labs and Fortune 100 companies
- [Customer Stories](https://www.v7darwin.com/customer-stories): Case studies across medical AI (Genmab, Intelligent Ultrasound, Vivan Therapeutics), automotive (Miovision, scaling automotive AI), agriculture (CattleEye), and infrastructure inspection (Abyss Solutions, Raptor Maps)
- [Security & Compliance](https://www.v7darwin.com/security): SOC 2 Type II and HIPAA compliance details, data handling policies, and enterprise security documentation

## Image & Video Annotation

- [Image Annotation](https://www.v7darwin.com/academy/annotations-getting-started): Bounding boxes, polygons, semantic segmentation, keypoints, and polylines with Auto-Annotate; label-similar-objects feature automatically finds matching instances to reduce repetitive manual effort across large datasets
- [Video Annotation](https://www.v7darwin.com/video-annotation): AI-assisted video labeling with SAM2 auto-track across frames, multi-camera sync, in/out-of-view detection, and support for all major video formats (.mp4, .mov, .mkv, .avi) at native frame rates; scales to 1,000+ annotations per frame and multi-hour videos

## Medical Imaging

- [Medical Imaging Annotation](https://www.v7darwin.com/medical-imaging-annotation): DICOM and NIfTI annotation with multiplanar (MPR) views, cinematic 3D rendering, MedSAM and TotalSegmentator auto-segmentation, whole slide imaging (WSI) support, and HIPAA-compliant infrastructure; reduces medical AI development time by up to 80%; customers include Roche, Boston Scientific, Genmab, and Vivan Therapeutics

## Data Labeling Services

- [Labeling Services](https://www.v7darwin.com/labeling-services): On-demand expert annotators for video, medical imaging (radiologists and pathologists), text and RLHF, defect inspection, agriculture, and retail; includes end-to-end project management from proof of concept through final delivery

## Developer Resources

- [Darwin API Reference](https://docs.v7labs.com/reference/introduction): REST API for programmatic access to datasets, workflows, and annotations — use when building custom integrations, automating data pipelines, or connecting Darwin to external tooling
- [Darwin Python SDK](https://darwin-py-sdk.v7labs.com/): Official `darwin-py` library (`pip install darwin-py`) for dataset management, data upload/download, PyTorch dataloader integration, and video frame extraction; supports Python 3.10–3.13
- [Darwin CLI Guide](https://docs.v7labs.com/docs/getting-started-1): Command-line interface for authenticating, creating and removing datasets, uploading images and videos, and exporting versioned releases

## Optional

- [Darwin Documentation](https://docs.v7labs.com/): Full docs index — getting started, annotation guides, dataset management, workflow design, external integrations, and billing
- [Darwin JSON Format](https://www.v7darwin.com/academy/darwin-json): Native annotation export/import schema; Darwin JSON 2.0 is the recommended format for DICOM and volumetric exports
- [Dataset Export Guide](https://www.v7darwin.com/academy/exports): Export annotations in TensorFlow, PyTorch, COCO, YOLO, Darwin JSON 2.0, and NIfTI formats
- [Loading Datasets in PyTorch](https://www.v7darwin.com/academy/pytorch-dataloaders): Guide for integrating Darwin datasets directly as PyTorch dataloaders for model training pipelines
- [Model Deployment API](https://docs.v7labs.com/docs/run-models-through-the-api): Deploy and serve trained computer vision models via REST API with code examples
- [HuggingFace Integration](https://docs.v7labs.com/docs/huggingface-integration): Connect HuggingFace models into Darwin workflows as model-in-the-loop pre-annotators
- [Changelog](https://www.v7darwin.com/changelog): Recent product releases — SAM 3 text-prompt segmentation, timeline mode for video, oblique views for medical imaging, 3D thresholding brush, and multi-camera sync