PP-OCRv5 mobile text detection (ONNX)
Text detector used by redact-web to read text in images, so personal information in screenshots and scans can be redacted, in the browser.
- Upstream: https://huggingface.co/PaddlePaddle/PP-OCRv5_mobile_det_onnx, at commit
e6f4fa85f00e168c862bc462aebca69eef9b3d3d, by PaddlePaddle (PaddleOCR). - License: Apache-2.0, the same as upstream (copy in
LICENSE). Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved. - Training data: not disclosed by PaddlePaddle. The PaddleOCR 3.0 report (arXiv:2507.05595) names no datasets and gives no data license.
Our modifications
None. tools/models/paddleocr/export.py downloads PaddlePaddle's official ONNX export, checks its SHA-256, and copies it byte for byte as onnx/model.onnx. onnx/inference.yml is the upstream file, unchanged.
Files
| File | Description |
|---|---|
onnx/model.onnx |
fp32 graph (upstream inference.onnx) |
onnx/inference.yml |
upstream pre- and post-processing config |
redact-web pins this repository at a commit and verifies every file's SHA-256 before use.