NTIRE 2025 Efficient Burst HDR and Restoration Challenge Dataset
收藏资源简介:
NTIRE 2025 Efficient Burst HDR and Restoration Challenge数据集是由CVLAI机构组织的一项挑战赛的数据集。该数据集旨在推动高效的多帧高动态范围(HDR)和恢复技术的进步。数据集包含每个场景九个具有不同曝光水平的噪声和错位的RAW帧。参赛者需要开发能够在严格效率约束下有效融合这些帧的解决方案。数据集包含了300个训练场景,每个场景包含九个RAW输入帧和一张地面真实RGB图像,以及20个验证场景。该数据集旨在鼓励参与者探索如何有效地利用多输入帧的信息,同时考虑每帧的独特特征。
The NTIRE 2025 Efficient Burst HDR and Restoration Challenge Dataset is a challenge dataset organized by the CVLAI institution. It aims to advance the development of efficient multi-frame high dynamic range (HDR) and image restoration technologies. Each scene in this dataset contains nine noisy and misaligned RAW frames with varying exposure levels. Participants are required to develop solutions that can effectively fuse these frames under strict efficiency constraints. The dataset comprises 300 training scenes, each including nine RAW input frames and one ground-truth RGB image, as well as 20 validation scenes. This dataset is designed to encourage participants to explore how to effectively utilize information from multiple input frames while considering the unique characteristics of each individual frame.

- 1NTIRE 2025 Challenge on Efficient Burst HDR and Restoration: Datasets, Methods, and ResultsCVLAI · 2025年



