WildDetect CLI Reference
Complete command-line interface reference for WildDetect.
Main Commands
detect
Run wildlife detection on images.
Arguments:
- images: Path to image file or directory
Options:
- -m, --model TEXT: Model path or MLflow model name
- -c, --config PATH: Configuration file path
- -o, --output PATH: Output directory
- --device TEXT: Device (cuda/cpu/auto)
- --batch-size INTEGER: Batch size
- --confidence FLOAT: Confidence threshold
- --tile-size INTEGER: Tile size for large images
- -v, --verbose: Verbose output
Examples:
# Basic detection
wildetect detect images/ --model detector.pt
# With config file
wildetect detect images/ -c config/detection.yaml
# Custom settings
wildetect detect images/ --model detector.pt --batch-size 32 --confidence 0.7
census
Run census campaign with analysis and reporting.
Arguments:
- campaign_name: Census campaign name
- images: Image directory
Options:
- -c, --config PATH: Configuration file (required)
- -o, --output PATH: Output directory
- --species TEXT: Target species (comma-separated)
- --generate-report: Generate PDF report
Examples:
analyze
Analyze detection results.
Arguments:
- results: Path to detection results JSON
Options:
- -o, --output PATH: Output directory
- --format TEXT: Output format (json/csv/excel)
fiftyone
Manage FiftyOne datasets.
Options:
- --action TEXT: Action (launch/info/export)
- --dataset TEXT: Dataset name
- --port INTEGER: Port number
Examples:
# Launch viewer
wildetect fiftyone --action launch --dataset my_dataset
# Get dataset info
wildetect fiftyone --action info --dataset my_dataset
# Export
wildetect fiftyone --action export --format coco --output export/
ui
Launch Streamlit web interface.
info
Show system and environment information.
For detailed script documentation, see WildDetect Scripts.