Crate visual_cortex_ocr_onnx 

PaddleOCR (detection + recognition) via ONNX Runtime, implementing visual-cortex-vision’s OcrEngine. Heavy dependencies are isolated here.

Structs

ModelFile

One pinned, checksummed model artifact.

PaddleOcr

PaddleOCR det+rec pipeline behind visual-cortex-vision’s OcrEngine.

Enums

OcrError

Constants

DET_MODEL

PP-OCRv4 mobile detection model (multilingual; finds text boxes).

REC_DICT

English recognition charset (95 entries; CTC classes = blank + 95 + space).

REC_MODEL

PP-OCRv3 English mobile recognition model (reads each box).

Functions

ensure_all_cached

Ensure all three artifacts are cached; returns (det, rec, dict) paths.