ESmike/true_imgs_modalities

该数据集包含图片及其相关属性，如图片的标题(caption)、尺寸(width和height)、文件名(filename)以及几种不同处理版本的图片哈希值(std512_hash、colorboost_hash、grayscale_hash和edges_hash)。图片本身包括边缘检测(edges)、灰度(grayscale)、标准512尺寸(std512)和颜色增强(colorboost)四种形式。数据集分为训练集(train)，包含4149个示例，整个数据集大小约为1.48GB。数据集适用于图像到图像、图像分割和图像到文本的任务。

The dataset includes images and their related attributes, such as image captions (caption), dimensions (width and height), filenames (filename), and hash values of several different processed versions of the images (std512_hash, colorboost_hash, grayscale_hash, and edges_hash). The images themselves include four forms: edge detection (edges), grayscale (grayscale), standard 512 size (std512), and color enhancement (colorboost). The dataset is split into a training set (train) with 4149 examples, and the entire dataset is approximately 1.48GB in size. The dataset is suitable for image-to-image, image segmentation, and image-to-text tasks.