- Uses MediaPipe Face Detection as primary detection method
- Implements dynamic confidence thresholds
- Optimized for both performance and accuracy
- Can be used exclusively in strict mode for higher accuracy
- Frontal face detection with adaptive scale factors
- Configurable detection parameters
- Enhanced with histogram equalization
- Only used in normal mode as fallback
- Specialized detection for side-facing portraits
- Automatic mirror detection for both left and right profiles
- Optimized for profile face characteristics
- Only used in normal mode as fallback
- Multi-stage detection pipeline
- Starts with MediaPipe for high accuracy
- Falls back to Haar Cascade for missed faces
- Additional profile face detection stage
- Maximum face detection coverage
- May detect non-human faces in some cases
- Uses only MediaPipe detection
- Higher accuracy for human faces
- Prevents false positives
- Suitable for professional portrait processing
- Recommended for applications requiring high precision
- Dynamic output size based on detected face dimensions
- Automatic head margin calculation
- Square aspect ratio preservation
- Face-centered composition
- Multi-stage preprocessing pipeline
- Contrast enhancement techniques
- Transparent background generation
- Anti-aliased circular mask option
- Supported formats: JPG, JPEG, PNG, WEBP
- Quality preservation system
- Automatic format handling
- PNG format with alpha channel
- Transparent background processing
- Optional circular mask with anti-aliasing
- Standardized output naming convention
- Efficient directory traversal
- Progress tracking system
- Detailed success/failure reporting
- Parallel processing capabilities