MOTHe

MOTHe detects and tracks multiple animals in heterogeneous video recordings using a convolutional neural network to quantify animal behavior and movement in natural habitats.

Key Features:

• End-to-End Pipeline: Provides an integrated computational pipeline from object detection to multi-animal tracking in video data.
• Convolutional Neural Network (CNN): Employs a basic CNN architecture for object detection to extract digital image features and classify animals in frames.
• Multi-Animal Tracking: Tracks multiple individuals across frames, including animals that are stationary or partially camouflaged in natural settings.
• Python Implementation: Implemented in Python as the computational environment for model execution and video processing.

Scientific Applications:

• Collective Movement Studies: Enables quantification of group-level movement patterns from video recordings.
• Animal Space-Use Analysis: Facilitates measurement of individual and group space-use metrics in natural habitats.
• Population Censuses: Supports automated counting and monitoring of animals in video-based surveys.
• Behavioral Neuroscience: Provides high-resolution behavioral tracking data for studies linking behavior to neural mechanisms.
• High-Resolution Behavioral Extraction: Extracts behavioral data from videos that are difficult to obtain via manual observation.

Methodology:

Python-based implementation using a basic Convolutional Neural Network (CNN) for object detection within an end-to-end detection-to-tracking pipeline that tracks multiple animals, including stationary or partially camouflaged individuals, in video recordings.