C# Tesseract Local OCR API

I need a small, self-hosted OCR service written in C# (.NET 6 or later) that accepts a Base64-encoded image and instantly returns the recognised text. Tesseract must be the engine under the hood, configured for English only, with any preprocessing kept lightweight so the full request/response cycle stays in the low-millisecond range on a typical workstation.

The DLL or minimal Web API you build has to be thread-safe so several parallel calls will not block each other. I’ll plug this straight into existing .NET applications, so please expose a clean public method or REST endpoint, document every parameter, and keep external dependencies to a minimum.

Deliverables
• Complete, well-commented source code and solution files
• Build/run instructions and NuGet package list
• A short demo project (console or unit test) showing a Base64 request and the returned text
• Any tuned Tesseract configs or traineddata you use

Acceptance criteria
The demo must extract English text from a sample image I provide with at least 95 % accuracy and respond in under 200 ms on my test machine. If these numbers slip we can profile and optimise together, but the baseline needs to be close.

Let me know your experience with Tesseract on .NET and any ideas you have for quick image pre-processing to boost accuracy without adding latency.