DrIFT: Autonomous Drone Dataset with Integrated Real and Synthetic Data, Flexible Views, and Transformed Domains
This addresses domain shift challenges for drone detection in airspace security, but it is incremental as it builds on existing dataset and adaptation methods.
The paper tackles the problem of visual drone detection being affected by domain shifts by introducing the DrIFT dataset with fourteen distinct domains and a new uncertainty-aware unsupervised domain adaptation method, achieving superior performance to state-of-the-art techniques.
Dependable visual drone detection is crucial for the secure integration of drones into the airspace. However, drone detection accuracy is significantly affected by domain shifts due to environmental changes, varied points of view, and background shifts. To address these challenges, we present the DrIFT dataset, specifically developed for visual drone detection under domain shifts. DrIFT includes fourteen distinct domains, each characterized by shifts in point of view, synthetic-to-real data, season, and adverse weather. DrIFT uniquely emphasizes background shift by providing background segmentation maps to enable background-wise metrics and evaluation. Our new uncertainty estimation metric, MCDO-map, features lower postprocessing complexity, surpassing traditional methods. We use the MCDO-map in our uncertainty-aware unsupervised domain adaptation method, demonstrating superior performance to SOTA unsupervised domain adaptation techniques. The dataset is available at: https://github.com/CARG-uOttawa/DrIFT.git.