asuglia/guesswhat-5k

该数据集是一个包含图像、问题和答案的多模态数据集，主要用于视觉问答（VQA）任务。数据集中的每个样本包含原始图像、图像的相关信息（如URL、文件名、尺寸等）、多个问题及其对应的答案、提问者ID、时间戳、对象ID、对话ID以及对象的详细信息（如边界框、类别、分割信息等）。数据集分为训练集、验证集和测试集，每个集合包含5000个样本。

This dataset is a multimodal dataset containing images, questions, and answers, primarily used for visual question answering (VQA) tasks. Each sample in the dataset includes the raw image, related information about the image (such as URL, file name, dimensions, etc.), multiple questions and their corresponding answers, questioner ID, timestamp, object ID, dialogue ID, and detailed information about objects (such as bounding boxes, categories, segmentation information, etc.). The dataset is divided into training, validation, and test sets, each containing 5000 samples.