Oztobuzz/tiny-gqa-bounding-box-detailed

这个数据集包含了问题、答案和与之相关的图片信息，每个样本都有一个问题ID、图片ID、问题文本、答案、关键对象ID、关键对象名称、对象边界框（包括高度、宽度、x坐标和y坐标）、问题类型（分为详细、语义和结构化类型）。数据集提供了一个训练集分割，共有18744个示例，大小为756,757,943字节。

The dataset consists of questions, answers, and related image information, with each sample having a question ID, image ID, question text, answer, key object ID, key object name, object bounding box (including height, width, x-coordinate, and y-coordinate), and question type (divided into detailed, semantic, and structural types). The dataset provides a training set split, containing 18,744 examples, with a size of 756,757,943 bytes.