Files
ssw c303c315e5 rocr: rusto-rs 源码入库(上游 v0.2.5 基线 57609fc + 本地补丁)
- 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
2026-09-12 17:30:04 +08:00

63 lines
1.9 KiB
Markdown

# 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
```bash
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:
- `# Title` for 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:
1. **Layout Model** (`layout.mnn`):
- Download from RapidDoc models or convert from PaddleOCR layout models
- Place in `models/DocOCR/`
2. **Detection Model** (`det.mnn`):
- Convert from PaddleOCR detection models using `convert_paddle_to_mnn.py`
3. **Recognition Model** (`rec.mnn`):
- Convert from PaddleOCR recognition models using `convert_paddle_to_mnn.py`
4. **Dictionary** (`dict.txt`):
- Download from PaddleOCR or RapidOCR repositories
See the main [README.md](../README.md) for model conversion instructions.