Deepdive Dataset (ROV multiclass image classification)

The dataset comprises 4158 images captured by a remotely operated vehicle (ROV), featuring deep-sea biota belonging to 62 different classes. Each image is meticulously labeled through rigorous quality control processes involving human-in-the-loop image labeling, ensuring accurate annotations.

本数据集包含4158张由遥控无人潜水器（Remotely Operated Vehicle，简称ROV）拍摄的图像，其内容为隶属于62个不同类别的深海生物。每张图像均通过包含人机协同图像标注环节的严格质量控制流程完成精细化标注，确保标注注释的准确性。