label-studio

Why we included this project

If you are building a machine learning pipeline, the quality of your training data usually matters more than the model architecture, and this is the tool teams reach for when they need to produce that data. Label Studio gives you a single web-based interface to annotate images, text, audio, video, and time series, with ready-made templates for object detection, semantic segmentation, named entity recognition, transcription, and more. It also covers the modern GenAI side of the job, supporting LLM fine-tuning data, RLHF preference collection, and side-by-side evaluation of model outputs. Because it exposes an API, Python SDK, and webhooks, you can wire it into an existing pipeline so predictions stream in for assisted labeling and finished annotations trigger training runs. That makes it a practical fit for small teams that want one flexible labeling platform rather than stitching together several point tools.