Drone Swarm Detection Dataset: Solving the Hardest Problem in Counter-UAS

The Saturation Problem

When a single drone approaches, detection is relatively straightforward. But what happens when fifteen drones arrive simultaneously from different directions? This is the saturation attack—and it's the scenario that breaks most Counter-UAS systems.

Current public drone detection datasets have a fundamental problem: they were built for yesterday's threats. Single consumer quadcopters, clean backgrounds, optimal lighting. Train on these datasets and your model will fail spectacularly when it encounters:

• Multiple targets appearing simultaneously (swarm tactics)
• Tiny objects at long range (10-50 pixel targets)
• Complex backgrounds (urban rubble, forests, smoke)
• Adverse weather (fog, rain, snow, low light)
• Non-standard drone shapes (FPVs, loitering munitions)

We built this dataset specifically to address these gaps.

What Makes This Dataset Different

Tiny Object Detection at Scale

One of the biggest challenges in Counter-UAS is detecting threats at range. At 500 meters, a small drone might only occupy 15-30 pixels in your optical sensor. Standard object detection models trained on consumer datasets simply aren't optimized for this.

Our dataset includes targets ranging from 10 to 50 pixels—the actual sizes you'll encounter in long-range surveillance scenarios. This lets you benchmark and improve your small object detection performance before deployment.

Dataset Structure

dataset/
 ├─ images/     # High-fidelity synthetic .jpg files
 └─ labels/     # YOLO format .txt annotations

Class Map:
ID | Class Name | Description
0  | drone      | Quadcopters, FPVs, loitering munitions

YOLO Configuration

# Simuletic Drone Swarm Configuration
path: /path/to/dataset
train: images
val: images

nc: 1
names: ['drone']

Use Cases

Download the Dataset

Ethics & License

Citation

@dataset{simuletic_drone_swarm_2025,
  author = {Simuletic Team},
  title = {Simuletic Synthetic Drone Swarm & Saturation Attack Dataset},
  year = {2025},
  url = {https://simuletic.com}
}