UAV Swarm: 12,598 Images for Object Detection and Tracking

12,598 annotated images of unmanned aerial vehicle swarms, preprocessed with a simulated monochrome near-infrared filter. The dataset contains 13 scenes and over 19 UAV types, with bounding box annotations in COCO format for a single 'uav' class. It was created by bnina-ayoub and last updated on Hugging Face in April 2026.

Use Cases

• Training object detection models based on the 12,598 annotated images with bounding boxes.
• Developing and benchmarking multiple object tracking algorithms based on the 13 sequential scenes.
• Simulating and testing sensor performance based on the CLAHE-enhanced, red-channel dominant monochrome NIR imagery.
• Researching swarm behavior and identification based on the over 19 different UAV types depicted.

Strengths

• 12,598 annotated images provide a substantial base for model training.
• Annotations are standardized in the widely-used COCO bounding box format [x, y, w, h].
• Includes 13 scenes and over 19 UAV types, suggesting diversity in scenarios and targets.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale training.
• The imagery is simulated and preprocessed with a specific NIR filter, which may not match real-world sensor data.

Provenance

Source

bnina-ayoub on Hugging Face

Collection Method

Images are preprocessed with a simulated Monochrome Near-Infrared (NIR) filter (CLAHE-enhanced, red-channel dominant channel mixing).

Time Range

null

Freshness

Last updated 2026-04-04 11:49:04; freshness should be verified.

Geography

null