Multi-view Drone Tracking Datasets

Overview

A collection of datasets for tracking drones using multiple camera views, enabling 3D trajectory reconstruction and multi-view tracking.

Key Features

• Multiple synchronized camera views
• 3D trajectory ground truth
• Camera calibration data
• Indoor and outdoor scenarios
• Suitable for multi-camera tracking systems

Dataset Components

The collection consists of 5 different datasets:

Dataset 1: Indoor Flight

• Indoor environment
• 3 synchronized cameras
• Controlled lighting conditions
• Precise camera calibration

Dataset 2: Outdoor Flight

• Outdoor environment
• 4 synchronized cameras
• Natural lighting conditions
• Longer flight trajectories

Dataset 3: Multi-Drone

• Multiple drones flying simultaneously
• 3 synchronized cameras
• Complex interaction scenarios
• Individual trajectory ground truth for each drone

Dataset 4: Occlusion Scenarios

• Specifically designed for occlusion handling
• 5 synchronized cameras
• Various obstacles creating occlusions
• Challenging tracking scenarios

Dataset 5: Winter Conditions

• Outdoor environment with snow
• 4 synchronized cameras
• Challenging weather conditions
• Low visibility scenarios

Strengths

• Unique multi-view perspective
• Precise 3D ground truth
• Supports advanced tracking algorithms
• Well-calibrated camera systems

Limitations

• Smaller scale than single-view datasets
• Limited environmental diversity
• Specialized equipment requirements
• Complex preprocessing needed

Source

GitHub: https://github.com/CenekAlbl/drone-tracking-datasets