LoLI-Street
收藏资源简介:
LoLI-Street数据集由韩国成均馆大学软件系创建,包含33,000张配对的低光和正常曝光的街道场景图像,涵盖19,000个对象类别,适用于低光图像增强和对象检测。数据集通过从不同城市收集的高分辨率视频中提取帧并手动筛选创建,确保图像质量。该数据集特别适用于自动驾驶和监控系统,旨在解决低光条件下图像增强和对象检测的挑战。
The LoLI-Street dataset was developed by the Department of Software, Sungkyunkwan University, Republic of Korea. It contains 33,000 paired low-light and normally exposed street scene images, covering 19,000 object categories, and is tailored for low-light image enhancement and object detection tasks. The dataset is constructed by extracting frames from high-resolution videos collected across various cities and performing manual filtering to ensure image quality. This dataset is particularly suitable for autonomous driving and surveillance systems, aiming to address the challenges of image enhancement and object detection under low-light conditions.

- 1LoLI-Street: Benchmarking Low-Light Image Enhancement and Beyond韩国成均馆大学软件系 · 2024年



