Label Studio is an open-source data labeling platform built by HumanSignal for annotating many data types with a single, standardized output format. Its defining trait is flexibility: labeling interfaces are defined through a configurable, XML-like template system, so the same tool can be adapted to tasks well beyond computer vision.
Supported modalities
It handles images, text, audio, video, and time-series data, and is commonly used for NLP, document AI, and audio transcription, as well as increasingly for LLM-oriented work such as prompt and response rating, RLHF, and model evaluation. Depending on how a labeling config is defined, annotators can draw boxes and polygons, classify and transcribe, tag named entities, or mark spans and regions.
Key features
- Configurable labeling templates for custom, multi-type tasks
- ML backend integration for model-assisted and active-learning workflows
- REST API, webhooks, and a standardized JSON output format
- Frontend distributed as an embeddable component for custom applications
Deployment and licensing
The Community Edition is free and open source under the Apache 2.0 license and can be self-hosted via pip or Docker. HumanSignal also offers managed cloud and enterprise editions that add team management, role-based access control, SSO, review workflows, and analytics for organizations that need governance and scale. Users range from individual researchers and small ML teams self-hosting the open-source app to enterprises standardizing annotation across mixed data types.
Multimodal