SynTIDE
收藏资源简介:
SynTIDE数据集是由华中科技大学的研究团队创建的,包含大约14K个水下场景的图像、深度图、语义掩模和对应的文本描述。该数据集通过TIDE模型生成,能够同时生成高度一致的水下场景图像和多种类型的密集注释,如深度图和语义掩模。数据集的构建旨在解决水下环境中密集预测任务的数据稀缺问题,为水下场景的深度估计和语义分割提供支持。
The SynTIDE dataset was created by a research team from Huazhong University of Science and Technology, containing approximately 14,000 underwater scene images, depth maps, semantic masks and corresponding textual descriptions. Generated via the TIDE model, this dataset can simultaneously produce highly consistent underwater scene images and multiple types of dense annotations such as depth maps and semantic masks. The dataset is constructed to address the issue of data scarcity for dense prediction tasks in underwater environments, providing support for depth estimation and semantic segmentation of underwater scenes.




