V7 Darwin Data Labeling Platform

V7 is an AI company founded in August 2018 by Alberto Rizzoli and Simon Edwardsson, headquartered in London with offices in San Francisco and New York. It builds tools that turn unstructured data into usable AI: V7 Darwin for training data and V7 Go for document workflows.

V7 Darwin is a labeling platform for images, video, PDFs, architectural drawings, DICOM and SVS pathology slides. Annotators work with AI assistance such as Auto-Annotate, SAM 2 and SAM 3 segmentation, auto-tracking across video and label-similar-objects search. Model-in-the-loop pre-labeling, multi-stage workflows with logic stages, consensus and labeler evaluation manage quality at scale. It runs as SaaS or in private cloud on AWS, Google Cloud and Azure, exposes an API and SDK, and is backed by managed labeling services for medical imaging, defect inspection, agriculture and retail.

• AI-assisted labeling: Auto-Annotate, SAM 2 and SAM 3 text-prompt segmentation.
• Video tools: auto-tracking, split and unify annotations, timeline summary and disagreement markers.
• Medical imaging: DICOM and SVS pathology slide support, with SAM 2 lesion segmentation on CT scans.
• Workflow builder: no-code multi-stage workflows with logic stages, roles and automations.
• Quality control: consensus stages, labeler evaluation and model-in-the-loop blind tests.
• Developer access: API, SDK, export to TensorFlow and PyTorch formats and private cloud deployment.

Hospitals and medical AI startups in Vietnam can label CT, MRI and pathology images for diagnostic models under HIPAA-grade controls. Electronics and garment factories can build defect inspection datasets from production line images. Vietnamese annotation outsourcing firms can run AI-assisted video labeling for foreign automotive and retail clients.

V7 states annotation up to 10x faster with AI-assisted labeling, and a nuclear industry customer, MTC, reported 9–10x faster labeling than its previous tool. Text-prompt segmentation with SAM 3 detects all instances of a class without manual drawing. Built-in consensus, labeler evaluation and review stages raise dataset quality, while SOC 2 Type II and HIPAA-compliant infrastructure allow sensitive medical data to be labeled.

Foundation segmentation models such as SAM are changing labeling from manual polygon drawing to prompt-and-review, cutting cost per image. Demand is rising for high-quality labeled data in medical imaging, industrial inspection and smart retail, and Vietnam has become a regional hub for data annotation outsourcing that needs efficient, auditable tooling.

Founded in August 2018; headquarters at 60 Margaret Street, London, with offices in San Francisco and New York. Raised a USD 3 million seed round (2020) and a USD 33 million Series A (2022). V7 Darwin listed on AWS Marketplace in March 2024; V7 Go launched in April 2024. SOC 2 Type II and HIPAA-compliant infrastructure.