高分辨率水下图像显著性目标检测训练数据集
收藏资源简介:
本数据集包含9641条样本,整体大小26.6GB,专门用于机器学习模型训练。每个样本包括: • RGB文件夹:存储高分辨率水下彩色图像(格式JPG),分辨率通常为1920x1080或更高,捕捉真实水下场景。 • masks文件夹:存储对应的二值掩码图像(格式PNG),用于标注显著性目标的精确区域(白色表示目标,黑色表示背景)。 • json文件夹:存储JSON文件,包含元数据和标注信息。 • labels文件夹:存储文本标签文件(格式TXT),适用于YOLO等目标检测框架,包含类别编号和归一化坐标。 数据集旨在支持水下图像显著性检测算法的开发,提升模型在复杂水下环境中的性能。
This dataset contains 9,641 samples with a total size of 26.6 GB, specifically designed for machine learning model training. Each sample includes the following components: • RGB folder: Stores high-resolution underwater color images in JPG format, typically with a resolution of 1920×1080 or higher, capturing realistic underwater scenes. • masks folder: Stores corresponding binary mask images in PNG format, used to annotate the precise regions of salient objects (white indicates the target, black indicates the background). • json folder: Stores JSON files containing metadata and annotation information. • labels folder: Stores text label files in TXT format, compatible with object detection frameworks such as YOLO, containing category IDs and normalized coordinates. This dataset is intended to support the development of underwater image salient object detection algorithms and improve the performance of models in complex underwater environments.




