PP-OCRv3 text detector (ONNX)
PaddleOCR's PP-OCRv3 text detector (DBNet), as ONNX. It finds where text is in an image; it doesn't read it. Card Anvil uses it in the browser to select text laid over card art.
These are unchanged copies of files from monkt/paddleocr-onnx at commit 7b02d0a, at the same paths, kept here so the app loads files that can't change under it.
| Files | What they are | Loaded by Card Anvil |
|---|---|---|
detection/v3/det.onnx, config.json |
The PP-OCRv3 text detector | Yes (det.onnx) |
detection/v5/det.onnx, config.json |
The PP-OCRv5 text detector: more accurate, 88 MB | No, kept for later |
languages/english/ |
A text reader (recognition model) for English, with its character dictionary | No, kept for later |
languages/latin/ |
A text reader for Latin-script languages, with its character dictionary | No, kept for later |
The source repo has more (other languages, document preprocessing); see it for those.
Credits and licence
- Model: PaddleOCR by PaddlePaddle.
- ONNX conversion:
monkt/paddleocr-onnx.
Licensed under the Apache License 2.0, as the source is. See LICENSE.