ssws3/Depth-Anything-V2-Datasets

Depth-Anything-V2数据集包含图像三元组，每个三元组由原始图像、灰度深度变体和着色深度变体组成。数据集用于深度估计或相关计算机视觉任务，提供基础图像ID、原始图像（如JPEG格式）以及对应的深度处理变体（可能为空）。示例展示了原始图像与深度图的对比，元数据包括图像尺寸、大小、URL和多语言描述（英语、印地语、Hinglish），以支持跨语言应用。该数据集旨在促进深度感知模型的研究和开发。The Depth-Anything-V2 dataset consists of image triplets, each comprising an original image, a greyscale depth variant, and a colourised depth variant. It is designed for depth estimation or related computer vision tasks, providing base image IDs, original images (e.g., in JPEG format), and corresponding depth-processed variants (which may be null). Examples illustrate comparisons between original images and depth maps, while metadata includes image dimensions, size, URLs, and multilingual captions (in English, Hindi, and Hinglish) to support cross-language applications. This dataset aims to facilitate research and development in depth-aware models.