Roboflow

Roboflow is a computer vision platform centered on the workflow from image data preparation to model deployment. The vendor site describes it as an end-to-end computer vision platform, while its deployment pages and public inference repository show support for hosted API access, dedicated deployments, self-hosting, and edge inference workflows.

What it does

• Organizes computer vision workflows around dataset preparation, training, and production deployment.
• Exposes deployment paths through web, API, and edge inference options.
• Provides Python-based inference tooling for hosted, dedicated, self-hosted, and edge use cases.

Deployment and integration surface

The Roboflow deploy page explicitly says production vision models can run on the web, via API, or on an edge inference device. The Python inference SDK documentation says the same SDK works with Roboflow-hosted API endpoints, dedicated deployments, self-hosted inference, and edge devices. The public roboflow/inference repository adds evidence for Docker-based and Python-driven inference workflows.

Why it fits the directory

Roboflow belongs in the Computer Vision category because the strongest supported evidence points to vision dataset workflows, model inference, and deployment tooling rather than general-purpose MLOps. The most defensible manufacturing relevance is for inspection, detection, and other camera-based automation use cases, but the identity of the product is broader than factory-only software.

Limitations

• The public GitHub metadata for roboflow/inference reports NOASSERTION for license, so the exact open-source licensing position of the inference repository is not confirmed from the metadata alone.
• The sources gathered in this session did not expose a fully detailed public pricing matrix, so pricing specifics remain incomplete.
• Evidence gathered here confirms deployment modes and inference tooling, but it does not yet provide a source-backed list of industrial integrations or protocol support beyond API-oriented access.