BREA-Depth
收藏资源简介:
BREA-Depth数据集由爱丁堡大学的Baillie Gifford Pandemic Science Hub和School of Informatics团队创建,包含3437张经过语义分割标注的支气管镜图像。数据集旨在解决支气管镜深度估计中的挑战,特别是提高复杂分支气道中的导航准确性和安全性。数据集的创建过程包括收集体外人肺模型的支气管镜视频数据,并通过深度感知CycleGAN框架进行训练,以提高模型的泛化能力和准确性。数据集适用于支气管镜导航和干预的3D气道重建等领域的研究。
The BREA-Depth dataset was developed by the team from the Baillie Gifford Pandemic Science Hub and the School of Informatics at the University of Edinburgh. It contains 3,437 bronchoscopic images annotated with semantic segmentation labels. The dataset is intended to address the challenges in bronchoscopic depth estimation, particularly improving the navigation accuracy and safety in complex branched airways. The development process of the dataset involves collecting bronchoscopic video data from ex vivo human lung models, followed by training with a depth-aware CycleGAN framework to enhance the generalization capability and accuracy of models built upon this dataset. This dataset is applicable to research in fields such as 3D airway reconstruction for bronchoscopic navigation and interventions.




