uscmcl/MCL_drone_dataset

A drone monitoring system integrating deep-learning-based detection and tracking modules is proposed in this work. The challenge in using deep learning for drone detection lies in the limited amount of training drone images. We address this by developing a model-based drone augmentation technique that generates drone images with bounding box labels. For tracking small flying drones, we utilize the residual information between consecutive image frames. Our integrated detection and tracking system outperforms individual modules containing only detection or tracking. Trained on synthetic data, the system performs admirably on real-world drone images against complex backgrounds. The USC drone detection and tracking dataset with user-labeled bounding boxes is publicly available.

Our USC drone dataset comprises 30 video clips, all captured on the USC campus using a single drone model. This dataset encapsulates a broad spectrum of backgrounds, camera angles, drone appearances, and weather conditions, aiming to represent the drone’s attributes in real-world scenarios like fast motion, extreme illumination, and occlusion.

Label Format: Each label is formatted as (x_center, y_center, width, height) in pixels.

Labeling Scheme: Starting from each folder, labels are provided for the first image out of every 15 images, corresponding to images like 0001.jpg, 0016.jpg, 0031.jpg, and so on.

Video Specifications: Each video is approximately one minute long, with a resolution of 1920 × 1080 and a frame rate of 15 frames per second.

More information: https://mcl.usc.edu/mcl-drone-dataset/