Annotate your AI data, as a team
Label text, images, audio, video, and structured data to train your models, thanks to Label Studio installed and maintained by DINAO. Your sensitive datasets remain on our French servers.
What is Label Studio?
Label Studio is an open-source data annotation and labeling platform designed for training and evaluating artificial intelligence models. It supports a wide variety of data : text (classification, entity recognition, relationships), images (bounding boxes, polygons, segmentation, OCR), audio (transcription, diarization), video (object tracking) as well as time series and tabular data.
Its strength lies in collaborative annotation: multiple people work on the same projects, with review, inter-annotator consensus, and agreement measurement. Coupled with a ML Backend, the platform automatically pre-annotates data using your own models, then lets your teams correct them — an active learning approach that significantly reduces labeling costs.
Label Studio integrates into MLOps pipelines via a REST API, webhooks, and exports in standard formats (COCO, YOLO, CoNLL, Pascal VOC, JSON), and connects to object storage (S3, MinIO, GCS, Azure). With DINAO, your often confidential datasets remain hosted in France.
Host Label Studio at DINAO
Resource tiers compatible with Label Studio prerequisites (minimum 2 cœurs / 8 Go / 10 Go). Hosted in France, fully managed.
- 4 dedicated vCPU
- 8 Go RAM
- 80 GB NVMe
- Daily backups
- Managed & monitored by DINAO
- 8 dedicated vCPU
- 16 Go RAM
- 160 GB NVMe
- Daily backups
- Managed & monitored by DINAO
2 hidden tier(s) (insufficient resources for this app) : Découverte, Standard
Technical details
You might be wondering…
What formats can I annotate?
Text, images, audio, video, time series, and tabular data. Label Studio covers classification, NER, image segmentation, transcription, video tracking, OCR, and many other tasks.
Can I accelerate annotation with AI?
Yes. Via an ML Backend, you connect a model that pre-labels the data; your annotators only need to correct it. Active learning prioritizes the most useful examples.
Are my datasets confidential?
Yes. Everything stays on your instance in France. Your training data, often sensitive, is never transmitted to a third party or used elsewhere.
How do I retrieve my annotations?
You export in standard formats (JSON, COCO, YOLO, CoNLL, Pascal VOC…) or via the API, to directly feed your training pipelines. No proprietary lock-in.
Where is the data hosted?
On DINAO infrastructure in France, in one of the available data centers. Your data never leaves the territory.
