LUMA
收藏资源简介:
LUMA数据集是由艾克斯-马赛大学、CNRS和LIS马赛联合创建的,旨在为处理不确定性和多模态数据提供一个基准。该数据集包含101,000张图像、135,096个音频记录和62,875个文本段落,总计约3GB。数据集通过Python工具包支持不确定性的控制注入,确保每种模态的数据与其对应的图像和音频记录相关联。LUMA数据集特别适用于需要处理多种数据类型和不同程度不确定性的深度学习模型,如医疗健康、自动驾驶和金融领域,以提高决策的准确性和可靠性。
The LUMA Dataset was jointly created by Aix-Marseille University, CNRS, and LIS Marseille, serving as a benchmark for uncertainty and multimodal data processing. This dataset comprises 101,000 images, 135,096 audio recordings, and 62,875 text passages, with a total size of approximately 3 GB. It supports controlled uncertainty injection via a Python toolkit, ensuring that data of each modality is correlated with its corresponding images and audio recordings. The LUMA Dataset is particularly suitable for deep learning models that need to handle multiple data types and varying degrees of uncertainty across fields such as healthcare, autonomous driving, and finance, to improve the accuracy and reliability of decision-making.




