sonson0910/engineering-drawing-detection

revert: restore clean baseline + update README with deploy links and technical report

fix: cv2.addWeighted dtype mismatch (float32 vs uint8) in draw_detections_enhanced

fix: NumpyEncoder for int64 JSON serialization + TTA toggle UI + updated README

fix: SyntaxError in overlap report formatter (nested f-string)

feat: implement proper overlap resolution per interviewer feedback

Restructure README to perfectly match the assessment requirements

Rewrite README to be more professional and remove emojis

Update README with actual web demo and model links

Translate README placeholders to English

Remove redundant debug files and scripts

Clean up redundant and testing comments

Pin numpy<2.0.0 to fix C-ABI breaking changes

Pin paddleocr==2.7.3 to fix v3.0 breaking changes

Fix PPStructure import error

Fix PaddleOCR Unknown argument show_log

Remove use_gpu argument to prevent unknown argument error in certain paddleocr versions

Fix torch.load missing explicit CPU mapping fallback and PaddleOCR GPU toggling on HF Space

Pin pydantic < 2.9.0

Upgrade Gradio to 5.x to fix Pydantic incompatibility

Pin huggingface-hub to 0.25.2 to fix HfFolder import error in Gradio

Update gradio sdk_version to fix huggingface_hub ImportError

Quote python_version to fix YAML float parsing as 3.1

Fix python version for HF Spaces

Upload model weights via LFS

Fix emoji in README

Fix HF Spaces configuration in README

Deploy to HF Spaces