M-MRE
收藏资源简介:
M-MRE数据集是基于现有GMNER数据集构建的,用于支持多模态信息提取领域的多模态相互强化效应(M-MRE)任务。该数据集融合了细粒度和粗粒度任务,包括实体识别、图像分割和图像描述。数据集通过利用大型语言模型生成图像描述,并结合现有的实体识别数据,实现了多模态任务的相互强化,旨在提升模型在多模态信息提取任务中的性能。
The M-MRE dataset is constructed based on the existing GMNER dataset, and is developed to support the multimodal mutual reinforcement effect (M-MRE) task in the field of multimodal information extraction. This dataset integrates both fine-grained and coarse-grained tasks, including entity recognition, image segmentation, and image captioning. By leveraging large language models to generate image captions and combining with existing entity recognition data, the dataset realizes mutual reinforcement among multimodal tasks, with the aim of improving the performance of models in multimodal information extraction tasks.

- 1M-MRE: Extending the Mutual Reinforcement Effect to Multimodal Information Extraction中国科学技术大学 · 2025年



