# Supervisely

> Computer vision platform for labeling images, video, 3D point clouds and medical scans, plus model training, evaluation and deployment.

Supervisely is a computer vision platform from Supervisely OÜ that covers data labeling, data curation, neural network training and deployment in one web-based environment. It supports images, videos, 3D point clouds (LiDAR) and DICOM medical imagery, and is offered as an online edition and a self-hosted Enterprise Edition. Its Python SDK is open source under the Apache 2.0 license.

## What It Is

Supervisely describes itself as an operating system for computer vision. Teams import data from local or cloud storage, annotate it with dedicated toolboxes for images, video, 3D point clouds and volumetric medical scans, and manage datasets with search, filtering and quality statistics. The same platform trains, evaluates and serves models, so labeled data can feed straight into model development.

## Labeling and Automation

The labeling tools include rectangle, polygon, skeleton and 3D lasso tools, nested ontologies, key-value tags, labeling jobs, reviews and quality metrics. Built-in Smart Tools based on SAM2, ClickSEG and others assist with segmentation. Video labeling offers auto-tracking and off-screen detection, and 3D labeling includes sensor fusion with synchronized 2D and 3D views. Users can also deploy custom or built-in models for auto labeling.

## Models, Data and Collaboration

The platform provides training infrastructure, model version comparison, performance reports and active learning workflows with a human in the loop. Data management connects to AWS, GCS and Azure storage without duplicating data, and offers embeddings plots, natural-language search and statistics such as class balance and co-occurrence heatmaps. Access control uses roles and teams, with OpenID, LDAP and MFA.

## Apps, SDK and Enterprise Deployment

Functionality is extended through Supervisely Apps, self-contained GUI extensions that can be public or private, with hundreds listed in an ecosystem catalog. Developers can integrate at five levels, from the HTTP REST API and Python SDK scripts to headless apps and apps embedded in labeling tools. The Enterprise Edition can be hosted on a company's own servers or cloud, including Kubernetes, and is installed and updated through a supervisely-cli command. The vendor states it is SOC2 aligned and GDPR compliant.

## Features
- Image labeling
- Video labeling with auto-tracking
- 3D point cloud and LiDAR labeling with sensor fusion
- DICOM and volumetric medical labeling
- AI-assisted Smart Tools (SAM2, ClickSEG)
- Auto labeling with built-in or custom models
- Labeling jobs, reviews and quality metrics
- Data management and natural-language search
- Dataset quality statistics and embeddings visualization
- Neural network training, evaluation and deployment
- Active learning workflows
- Synthetic data generation
- Supervisely Apps ecosystem
- REST API and Python SDK
- Self-hosted Enterprise Edition with SSO and role-based access
- Managed labeling workforce services

## Integrations
AWS S3, Google Cloud Storage, Azure, LDAP, Microsoft Active Directory, OAuth2, OpenID, Kubernetes, AWS EKS, Docker, GitHub, GitLab, Slack

## Platforms
WINDOWS, MACOS, LINUX, WEB, API, VSC_EXTENSION, DEVELOPER_SDK, CLI

## Pricing
Freemium — Free tier available with paid upgrades

## Version
v6.74.48

## Links
- Website: https://supervisely.com
- Documentation: https://docs.supervisely.com/
- Repository: https://github.com/supervisely/supervisely
- EveryDev.ai: https://www.everydev.ai/tools/supervisely