from Hacker News

We're excited to share florence2-sharp, a C# library implementing the Florence-2-model for advanced image understanding tasks. Florence-2 uses a prompt-based approach to a variety of vision tasks, and provides great zero-shot performance across many vision tasks.

Our C# library supports:

- Image captioning (from concise to detailed)

- Optical Character Recognition (OCR)

- Region-based OCR

- Object detection

- Optional phrase grounding

The library is a C# port of Microsoft's Florence-2 model (from https://huggingface.co/microsoft/Florence-2-base), based on the original model and the JS port by Frank Krueger (https://github.com/praeclarum/transformers-js).

Repo: https://github.com/curiosity-ai/florence2-sharp

NuGet: https://www.nuget.org/packages/Florence2