Movie Posters DataSet (MPDS)
收藏资源简介:
MPDS数据集是由南京理工大学和SeetaCloud合作创建的,专门用于电影海报生成的图像-文本对数据集。该数据集包含超过373,000对图像-文本对和8,000多张演员图像,覆盖了4,000多名演员。数据集的创建过程结合了自动化的视觉语言模型和人工校正,确保了描述的准确性和完整性。MPDS数据集的应用领域主要集中在个性化电影海报生成,旨在通过先进的扩散模型技术提升电影海报的设计效率和质量。
The MPDS dataset is a specialized image-text pair dataset for movie poster generation, co-developed by Nanjing University of Science and Technology and SeetaCloud. It contains over 373,000 image-text pairs, as well as more than 8,000 actor images covering over 4,000 distinct actors. The construction of the dataset integrates automated vision-language models and manual correction, which guarantees the accuracy and completeness of its descriptions. The primary application scenario of the MPDS dataset is personalized movie poster generation, where it aims to enhance the design efficiency and quality of movie posters through advanced diffusion model technologies.




