This project provides a comprehensive system for detecting bullet holes in archery target images using a multi-stage image processing pipeline.

1. Clone the repository:

git clone https://github.com/your-repo/archery-target-detection.git

2. Navigate to the project directory:

cd archery-target-detection

3. Create a virtual environment and activate it:

python -m venv hole_det_venv
source hole_det_venv/bin/activate

4. Install the required dependencies:

pip install -r requirements.txt

The project includes a comprehensive processing pipeline implemented in process_pipeline.py. This script chains multiple image processing stages together.

python process_pipeline.py input.jpg

python process_pipeline.py input.jpg \
 --edge-output edges_custom.png \
 --clean-output clean_custom.png \
 --blob-output final_blobs.png

python process_pipeline.py input.jpg \
 --blur-size7 7 \
 --canny-low 100 \
 --canny-high 200

python process_pipeline.py input.jpg \
 --blob-mode filter \
 --min-area 50 \
 --max-area 5000

python process_pipeline.py input.jpg \
 --keep-intermediate \
 --blob-output results.png

For detailed help and all available options:

python process_pipeline.py --help

The detection system uses a multi-stage approach:

1. Preprocessing: Enhances image quality
2. Multi-scale Detection: Uses Hough Circle Transform and contour analysis
3. Target Ring Filtering: Eliminates false positives from target scoring rings
4. Validation: Filters results based on confidence, darkness, and overlap

{
 "gamma_correction": 1.2,
 "clahe_clip_limit": 4.0,
 "clahe_grid_size": [8,8],
 "morph_kernel_size": 2,
 "morph_iterations": 2,
 "blur_kernel": [3,3],
 "blur_sigma": 0
}

{
 "param1": 25,
 "param2": 8,
 "min_radius": 5,
 "max_radius": 50,
 "accum_threshold": 15,
 "max_accu": 30
}

{
 "darkness_threshold": 85,
 "confidence_threshold": 0.3,
 "overlap_threshold": 0.7,
 "size_variation_threshold": 0.5,
 "min_hole_intensity": 35,
 "max_hole_intensity": 125
}

• Increase min_hole_intensity to 30-35
• Increase max_hole_intensity to 125-130
• Decrease darkness_threshold to 80-85

• Decrease min_hole_intensity to 40-45
• Decrease max_hole_intensity to 115-120
• Increase darkness_threshold to 85-90

1. Too few holes detected: Check darkness_threshold, min_radius, max_radius
2. Too many false positives: Increase confidence_threshold, decrease target_ring_tolerance

• Primary Method: Hough Circle Transform (multi-scale)
• Secondary Method: Contour analysis for edge cases
• Validation: Circularity, darkness, and spatial filtering

For issues or questions:

1. Check parameter tuning guide
2. Review results.json for detailed detection information
3. Consider image quality and lighting conditions

1. Use high-resolution images (minimum 1000x1000 pixels)
2. Ensure even lighting across the target
3. Adjust parameters based on target material and color
4. Review results.json for accuracy
5. For batch processing, consider parallel processing