Pipeline Overview¶
Figure: KABR tools pipeline for processing drone videos. The pipeline consists of four main steps: video data collection, data pre-processing, behavior labeling, and ecological analysis. Each step is modular and can be adapted to different species and study systems.
Pipeline Steps¶
The KABR tools pipeline consists of four main sequential steps:
1. Video Data Collection¶
Collection of drone video footage following best practices for wildlife observation while minimizing disturbance to animals.
2. Data Pre-processing with CVAT¶
Creation of mini-scenes focused on individual animals using detection and tracking, either manual or automated approaches.
3. Behavior Labeling¶
Application of machine learning models to classify animal behaviors from the processed mini-scenes.
4. Ecological Analysis¶
Generation of ecological insights including time budgets, behavioral transitions, social interactions, and spatial analysis.
Key Features¶
- Modular Design: Each step can be adapted and customized for different species and study systems.
- Multiple Detection Options: Support for both manual annotation and automated YOLO-based detection.
- Machine Learning Integration: Pre-trained models available with support for custom model training.
- Comprehensive Analytics: Generate multiple types of ecological metrics and visualizations.
Getting Started¶
Follow the pipeline steps in order:
Pre-processed example data for each step and pre-trained models are available on Hugging Face to help you get started quickly.
For additional customization and advanced features, see Optional Steps.