bnina-ayoub/UAV-Swarm

---
license: cc-by-4.0
task_categories:
- object-detection
- multiple-object-tracking
tags:
- uav
- drone
- aerial
- detection
- tracking
pretty_name: UAVSwarm
dataset_info:
features:
- name: image_id
dtype: int64
- name: image
dtype: image
- name: width
dtype: int32
- name: height
dtype: int32
- name: objects
struct:
- name: id
list: int64
- name: area
list: float32
- name: bbox
list:
list: float32
length: 4
- name: category
list:
class_label:
names:
'0': uav
splits:
- name: train
num_bytes: 1126114608
num_examples: 13188
- name: validation
num_bytes: 526847354
num_examples: 6288
- name: test
num_bytes: 900593061
num_examples: 11008
download_size: 2553674483
dataset_size: 2553555023
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
---

# UAVSwarm Dataset

Unmanned Aerial Vehicle Swarm dataset for multiple object tracking and detection.
Images are preprocessed with a simulated **Monochrome Near-Infrared (NIR)** filter
(CLAHE-enhanced, red-channel dominant channel mixing).

- 13 scenes, 19+ UAV types, 12,598 annotated images
- Single class: `uav` (id=0)
- Bounding boxes in COCO format `[x, y, w, h]`

## Load

```python
from datasets import load_dataset
dataset = load_dataset("bnina-ayoub/UAV-Swarm")

categories = dataset["train"].features["objects"].feature["category"].names
id2label = {i: c for i, c in enumerate(categories)}
```

## Citation
```bibtex
@article{wang2022uavswarm,
title = {UAVSwarm Dataset: An Unmanned Aerial Vehicle Swarm Dataset for Multiple Object Tracking},
author = {Wang, C. and Su, Y. and Wang, J. and Wang, T. and Gao, Q.},
journal = {Remote Sensing},
volume = {14},
number = {11},
pages = {2601},
year = {2022},
doi = {10.3390/rs14112601}
}
```