VietCat/TrafficSignDetector

Add stdout flush to ensure real-time log output

Add detailed logging for inference timing and top detections

Add input validation to prevent None image errors

Fix tiling strategy to avoid infinite loops

Implement tiling strategy for high-resolution object detection

Remove zoom feature - keep interface clean and simple

Simplify zoom to bare minimum - Gradio handles scroll natively

Simplify zoom - use Image components with native scroll

Fix horizontal scroll - remove Column constraints and add white-space nowrap

Fix horizontal zoom stretching - allow width to scale properly

Replace zoom with scrollable HTML display - actual pixel-level zoom

Fix reset button output error - remove None from outputs list

Simplify zoom implementation using global variables

Fix zoom slider functionality - now updates image display directly

Add zoom sliders for output and preprocessed images

Add auto-detection, interactive zoom, and class names to top 5 detections

Add confidence threshold slider and update class labels

Add detailed training prompt and guidelines

Lower confidence threshold to 0.001 and add detailed confidence analysis

Add raw predictions debug info and scale factor warnings

Add detailed logging for debugging detection pipeline

Add full logging and preprocessed image visualization

Fix model loading from HuggingFace Hub and improve inference preprocessing

remove test existing model

test existing model

fix bug

fix bug

fix bug

fix bug

fix bug

fix bug

add more log to debug

add more logic to process image

Lower confidence threshold to 0.25 and add debug logging for traffic sign detection

add more debug log

Use monkey patch to disable weights_only for ultralytics model loading

Add DetectionModel to safe globals for torch loading

Add torch safe globals for model loading

Fix repo_id parsing for HF Hub download

Download model from HF Hub at runtime

Force reinstall numpy 1.24.3 after all packages

Downgrade numpy to 1.24.3 for compatibility

Force reinstall opencv-python-headless to override full opencv

Downgrade gradio to 3.50.2 for compatibility

Add huggingface-hub version for compatibility

Load model at runtime, specify full path to model file

Use opencv-python-headless to avoid GUI dependencies

Add system dependencies for OpenCV in Docker

Initial commit with Traffic Sign Detector app

initial commit