ESHDR
收藏资源简介:
ESHDR数据集是由上海人工智能实验室、浙江大学、香港中文大学和华中科技大学联合开发的,专门用于12-stop HDR成像的第一个带有同步事件信号的数据集。该数据集通过一个新颖的仿真管道生成,包含了从-6EV到+6EV的不同曝光级别的LDR图像,并结合了高动态范围的事件信号。数据集的创建旨在解决动态场景中高动态范围成像的挑战,特别是图像对齐和融合问题。ESHDR数据集的应用领域主要集中在计算摄影中的HDR成像,旨在提高动态场景中HDR成像的质量和精度。
The ESHDR dataset was jointly developed by the Shanghai AI Laboratory, Zhejiang University, The Chinese University of Hong Kong, and Huazhong University of Science and Technology. It is the first dataset with synchronized event signals specifically designed for 12-stop HDR imaging. Generated via a novel simulation pipeline, the dataset contains LDR images with exposure levels ranging from -6EV to +6EV, combined with high dynamic range event signals. The dataset was created to address the challenges of high dynamic range imaging in dynamic scenes, particularly image alignment and fusion issues. The application fields of the ESHDR dataset mainly focus on HDR imaging in computational photography, with the goal of improving the quality and accuracy of HDR imaging in dynamic scenes.

- 1Event-assisted 12-stop HDR Imaging of Dynamic Scene上海人工智能实验室、浙江大学、香港中文大学、华中科技大学 · 2024年



