- 源码 316 文件收编为 rusto-rs/ 文件夹,随本仓库管理 - 移除内部 .git(补丁已存档 docs/rusto-rs-metal-patch-20260912.patch,上游可随时 re-clone) - target/ 与 docs/.yarn 缓存不入库;本地改动:ROCR_* 环境变量调优 + Metal 链接修复 + 预处理测试脚本
1.9 KiB
1.9 KiB
Examples
This directory contains example applications demonstrating various features of RustO!
Document Pipeline Demo
File: doc_pipeline_demo.rs
Demonstrates the full document processing pipeline with layout analysis and OCR.
Features
- Layout detection using PP-DocLayout models from RapidDoc
- Text recognition using PaddleOCR models
- Markdown output generation
- Support for multiple layout element types
Usage
cargo run --example doc_pipeline_demo -- \
--image path/to/document.jpg \
--layout-model models/DocOCR/layout.mnn \
--det-model models/ch_PP-OCRv4_det_infer.mnn \
--rec-model models/ch_PP-OCRv4_rec_infer.mnn \
--keys-path models/ppocr_keys_v1.txt
Arguments
--image: Path to the input document image--layout-model: Path to the layout detection model (default:models/DocOCR/layout.mnn)--det-model: Path to the text detection model (default:models/ch_PP-OCRv4_det_infer.mnn)--rec-model: Path to the text recognition model (default:models/ch_PP-OCRv4_rec_infer.mnn)--keys-path: Path to the character dictionary (default:models/ppocr_keys_v1.txt)
Output
The example generates markdown-formatted output with:
# Titlefor titles**Header**for headers*Caption*for figure/table captions- Plain text for body text
- Placeholder indicators for figures and tables
Required Models
You need to download or convert the following models:
-
Layout Model (
layout.mnn):- Download from RapidDoc models or convert from PaddleOCR layout models
- Place in
models/DocOCR/
-
Detection Model (
det.mnn):- Convert from PaddleOCR detection models using
convert_paddle_to_mnn.py
- Convert from PaddleOCR detection models using
-
Recognition Model (
rec.mnn):- Convert from PaddleOCR recognition models using
convert_paddle_to_mnn.py
- Convert from PaddleOCR recognition models using
-
Dictionary (
dict.txt):- Download from PaddleOCR or RapidOCR repositories
See the main README.md for model conversion instructions.