Skip to content

WildDetect Monorepo Documentation

WildDetect

Welcome to the comprehensive documentation for the WildDetect project - a complete wildlife monitoring and conservation toolkit for aerial imagery analysis.

What is WildDetect?

WildDetect is an integrated ecosystem of three specialized packages designed to streamline the entire wildlife detection workflow, from data management to model training and deployment:

  • WilData - Data pipeline and management
  • WildTrain - Model training and evaluation
  • WildDetect - Detection deployment and census analysis

Key Features

🎯 Complete Wildlife Detection Pipeline

  • Multi-species detection using state-of-the-art YOLO models
  • Batch processing of large-scale aerial imagery
  • Automated census campaigns with population statistics
  • Geographic visualization and analysis

📦 Comprehensive Data Management

  • Import from COCO, YOLO, and Label Studio formats
  • Data transformations (tiling, augmentation, filtering)
  • DVC integration for dataset versioning
  • ROI extraction for hard sample mining

🧠 Flexible Training Framework

  • Support for YOLO and MMDetection frameworks
  • Classification and object detection training
  • MLflow experiment tracking
  • Hyperparameter optimization with Optuna

🌍 Geographic Analysis

  • GPS metadata extraction and management
  • Flight path analysis and coverage maps
  • Interactive visualizations with FiftyOne
  • Population density and distribution analysis

Quick Navigation

For New Users

Start here to get up and running quickly:

For Researchers & Conservationists

Learn how to use the tools for your wildlife monitoring needs:

For Developers

Understand the architecture and extend the toolkit:

For Data Scientists & ML Engineers

Prepare datasets and train models:

Package Overview

🗂️ WilData - Data Pipeline

The foundation for dataset management and preparation.

Key Capabilities: - Import datasets from multiple formats (COCO, YOLO, Label Studio) - Apply transformations (tiling, augmentation, bbox clipping) - Create ROI datasets for classification - Update GPS metadata from CSV files - DVC integration for version control - REST API for programmatic access

Learn more about WilData →

🎓 WildTrain - Training Framework

Modular training system for detection and classification models.

Key Capabilities: - YOLO and MMDetection framework support - PyTorch Lightning for classification - Hydra configuration management - MLflow experiment tracking - Model registration and versioning - Hyperparameter tuning

Learn more about WildTrain →

🔍 WildDetect - Detection & Analysis

Production-ready detection and census system.

Key Capabilities: - Multi-threaded detection pipelines - Raster (large image) detection support - Census campaign orchestration - Geographic analysis and visualization - FiftyOne integration - Comprehensive reporting (JSON, CSV)

Learn more about WildDetect →

Common Workflows

Detection Workflow

graph LR
    A[Aerial Images] --> B[WildDetect]
    B --> C[Detections]
    C --> D[Analysis]
    D --> E[Reports & Maps]

Training Workflow

graph LR
    A[Annotations] --> B[WilData]
    B --> C[Processed Dataset]
    C --> D[WildTrain]
    D --> E[Trained Model]
    E --> F[WildDetect]

Census Workflow

graph LR
    A[Flight Planning] --> B[Image Capture]
    B --> C[WildDetect]
    C --> D[Statistics]
    D --> E[Geographic Viz]
    E --> F[Reports & Maps]

Getting Help

  • Tutorials: Step-by-step guides for common tasks
  • API Reference: Complete command and function documentation
  • Troubleshooting: Solutions to common issues
  • GitHub Issues: Report bugs or request features

System Requirements

  • Python: 3.9 or higher
  • GPU: CUDA-capable GPU recommended (optional)
  • OS: Windows, Linux, macOS
  • Memory: 16GB RAM minimum, 32GB recommended
  • Storage: SSD recommended for large datasets

Contributing

We welcome contributions! This is an open-source project designed for the conservation community.

  • Submit bug reports and feature requests via GitHub Issues
  • Contribute code via pull requests
  • Share your use cases and results
  • Help improve documentation

License

MIT License - see LICENSE files in each package for details.

Citation

If you use WildDetect in your research, please cite:

@software{wildetect2025,
  author = {Seydou Fadel M., Allin Paul},
  title = {WildDetect: Wildlife Detection and Census System for Aerial Imagery},
  year = {2025},
  url = {https://github.com/fadelmamar/wildetect}
}

Support

For questions and support: - 📧 Email: [fadel.seydou{at}delcaux{dot}com] - 💬 GitHub Discussions - 🐛 GitHub Issues

Ready to get started? Head to the Installation Guide to set up your environment.