Description
OCRing hundreds of thousands of scanned pages for RAG takes days at a few pages per second. TurboOCR is an ultra-fast OCR inference server on TensorRT FP16 and PP-OCRv6, processing 200+ images per second on OmniDocBench.
It offers HTTP and gRPC APIs with document parsing and PDF extraction for RAG pipelines.
Fast:200+ images/s.
TensorRT:FP16.
APIs:HTTP and gRPC.
Documents:PDF extraction.
It offers HTTP and gRPC APIs with document parsing and PDF extraction for RAG pipelines.
Features
Fast:200+ images/s.
TensorRT:FP16.
APIs:HTTP and gRPC.
Documents:PDF extraction.
