Chinese text from a picture: third-party files in this folder ============================================================== Everything the reader needs is served from this folder. Nothing is fetched from another host, and the picture a visitor chooses never leaves the browser. 1. ONNX Runtime Web 1.30.0 (the runtime that executes the models, as WebAssembly) Files: ort.wasm.min.js, ort-wasm-simd-threaded.mjs, ort-wasm-simd-threaded.wasm Source: npm package "onnxruntime-web", version 1.30.0, folder dist/, unmodified https://github.com/microsoft/onnxruntime Licence: MIT, Copyright (c) Microsoft Corporation. See LICENSE-onnxruntime.txt and ThirdPartyNotices-onnxruntime.txt. 2. PP-OCR models (PaddleOCR) Files: PP-OCRv6_det_tiny.onnx text detection (PP-OCRv6, tiny) PP-OCRv6_rec_tiny.onnx text recognition (PP-OCRv6, tiny), used for Simplified ch_PP-OCRv5_rec_mobile.onnx text recognition (PP-OCRv5, mobile), used for Traditional ppocrv6_tiny_dict.txt character list of PP-OCRv6_rec_tiny ppocrv5_dict.txt character list of ch_PP-OCRv5_rec_mobile Source: PaddleOCR, https://github.com/PaddlePaddle/PaddleOCR The .onnx files are the conversions published by the RapidOCR project (https://github.com/RapidAI/RapidOCR, release v3.10.0), unmodified. Notice: Converted from official PaddleOCR models to the ONNX format by the RapidOCR project. The model weights are copyright Baidu and/or the applicable PaddleOCR rights holders and are licensed under the Apache License, Version 2.0. The conversion and packaging are copyright the RapidOCR Authors, also under the Apache License, Version 2.0. Licence: Apache License 2.0. See LICENSE-paddleocr.txt. SHA-256: f42c0fbd294d95eac1a550e131b277dac97462c8025fa4b6c3cec1b7894bd3d5 PP-OCRv6_det_tiny.onnx e16e242de5937ad92609223f19bc2aff3727ee40b095f996907c24749bad251b PP-OCRv6_rec_tiny.onnx 5825fc7ebf84ae7a412be049820b4d86d77620f204a041697b0494669b1742c5 ch_PP-OCRv5_rec_mobile.onnx The names PaddleOCR, PaddlePaddle, RapidOCR and ONNX Runtime are used here only to say where the files come from. No endorsement by their owners is implied. To upgrade: take the new files from the same sources, keep the file names (or change them in FILES at the top of ocr.js together with the sizes), and update the versions and hashes above.