infinite-dataset-hub/Corroboration-Image

--- license: mit tags: - infinite-dataset-hub - synthetic --- # Corroboration-Image tags: multimodal, visual verification, image corroboration _Note: This is an AI-generated dataset so its content may be inaccurate or false_ **Dataset Description:** The 'Corroboration-Image' dataset is a curated collection of images paired with textual evidence used for training machine learning models in the task of visual verification and image corroboration. The dataset is designed to support multimodal learning approaches, where a model learns to associate textual descriptions with corresponding visual content to confirm or deny the validity of a claim or statement. The labeling strategy focuses on 'Corroboration' to indicate whether the visual content supports the textual claim provided. **CSV Content Preview:** ``` id, image_url, claim, label 1, "http://example.com/images/corroboration1.jpg", "A group of people are seen at a concert, loudly cheering.", "Corroboration" 2, "http://example.com/images/corroboration2.jpg", "The picture shows a green car parked in front of a red building.", "Corroboration" 3, "http://example.com/images/corroboration3.jpg", "An image depicts a historical monument with a plaque detailing its history.", "Corroboration" 4, "http://example.com/images/corroboration4.jpg", "This photograph captures a cat playing with a ball of yarn.", "Corroboration" 5, "http://example.com/images/corroboration5.jpg", "A snapshot shows a variety of fruits on a farm in an orchard.", "Corroboration" ``` In this dataset, the 'id' column represents a unique identifier for each image-text pair, 'image_url' contains the URL to the corresponding image, 'claim' includes a textual description that should be corroborated by the image, and 'label' is either 'Corroboration' or 'No Corroboration' to indicate if the image confirms the textual claim. **Source of the data:** The dataset was generated using the [Infinite Dataset Hub](https://huggingface.co/spaces/infinite-dataset-hub/infinite-dataset-hub) and microsoft/Phi-3-mini-4k-instruct using the query 'corroboration ': - **Dataset Generation Page**: https://huggingface.co/spaces/infinite-dataset-hub/infinite-dataset-hub?q=corroboration+&dataset=Corroboration-Image&tags=multimodal,+visual+verification,+image+corroboration - **Model**: https://huggingface.co/microsoft/Phi-3-mini-4k-instruct - **More Datasets**: https://huggingface.co/datasets?other=infinite-dataset-hub--- 许可证：MIT协议 标签： - 无限数据集中心（Infinite Dataset Hub） - 合成数据集 --- # 佐证图像（Corroboration-Image） 标签：多模态、视觉验证、图像佐证 注意：本数据集由人工智能生成，其内容可能存在不准确或虚假情况。 **数据集描述：** "佐证图像（Corroboration-Image）"数据集是经精心整理的图像与文本证据配对集合，用于训练机器学习模型完成视觉验证与图像佐证任务。本数据集旨在支持多模态学习范式，即模型通过学习将文本描述与对应视觉内容建立关联，以验证或驳斥某一主张或表述的真实性。其标注策略围绕"佐证"展开，用于表明视觉内容是否支持给定的文本主张。 **CSV内容预览：** id, image_url, claim, label 1, "http://example.com/images/corroboration1.jpg", "A group of people are seen at a concert, loudly cheering.", "Corroboration" 2, "http://example.com/images/corroboration2.jpg", "The picture shows a green car parked in front of a red building.", "Corroboration" 3, "http://example.com/images/corroboration3.jpg", "An image depicts a historical monument with a plaque detailing its history.", "Corroboration" 4, "http://example.com/images/corroboration4.jpg", "This photograph captures a cat playing with a ball of yarn.", "Corroboration" 5, "http://example.com/images/corroboration5.jpg", "A snapshot shows a variety of fruits on a farm in an orchard.", "Corroboration" 本数据集中，"唯一标识符（id）"列代表每个图像-文本配对的唯一标识，"图像链接（image_url）"字段存储对应图像的访问地址，"主张文本（claim）"字段包含需由图像佐证的文本描述，"标注标签（label）"仅可取"Corroboration"或"No Corroboration"两种取值，用于表明图像是否确认该主张文本的真实性。 **数据集来源：** 本数据集通过[无限数据集中心（Infinite Dataset Hub）]与微软（Microsoft）的Phi-3-mini-4k-instruct模型，基于查询词"corroboration"生成。 - **数据集生成页面**：https://huggingface.co/spaces/infinite-dataset-hub/infinite-dataset-hub?q=corroboration+&dataset=Corroboration-Image&tags=multimodal,+visual+verification,+image+corroboration - **所用模型**：https://huggingface.co/microsoft/Phi-3-mini-4k-instruct - **更多数据集**：https://huggingface.co/datasets?other=infinite-dataset-hub