HuaweiOCR

English | 中文

Windows-first batch OCR pipeline for device labels.

Quick Start

python -m pip install -r requirements.txt

Create .env:

API_KEY=your_api_key_here

Put images in new_images, then run:

python run_all.py --input new_images --out runs --format jsonl --log-level info --device cpu

Or double-click start.bat. The script creates new_images if needed and stops with a clear prompt when the folder is empty.

Outputs

Results are written under runs/. If output folders already exist, a per-run sibling folder is created to avoid overwrites.

• stage1_labels/ or stage1_labels_run_*
• stage2_fields/model/
• stage2_fields/sn/
• stage2_fields/manifest.jsonl
• stage2_fields/model_sn_ocr.jsonl
• stage2_fields/debug_ocr_barcode.log only with --log-level debug

See README_EN.md or README_ZH.md for details.