GoPro Random Sharp (GoProRS)
收藏资源简介:
GoProRS数据集是由哈尔滨工程大学智能系统科学与工程学院、卡利亚里大学和圣安娜高等研究学院联合创建的视频去模糊数据集。该数据集基于GoPro数据集生成,包含33个视频序列,每个序列中的清晰帧比例可由用户自定义,从而增强了模型的泛化能力。数据集通过高帧率摄像机捕捉的1280×720分辨率视频生成,帧率为250fps。GoProRS的创建旨在解决现有数据集中清晰帧比例固定的问题,使其更适用于真实世界的动态模糊场景。该数据集广泛应用于视频去模糊领域,旨在提升自动驾驶、面部识别和安防监控等应用的视频质量。
The GoProRS dataset is a video deblurring dataset jointly created by the School of Intelligent Systems Science and Engineering of Harbin Engineering University, the University of Cagliari and Scuola Superiore Sant'Anna. Constructed based on the GoPro dataset, it contains 33 video sequences, where the ratio of sharp frames in each sequence can be customized by users, thus enhancing the generalization ability of models. The dataset is generated from videos captured by high-frame-rate cameras at a resolution of 1280×720, with a frame rate of 250 fps. The GoProRS dataset was developed to address the issue of fixed sharp frame ratio in existing datasets, making it more applicable to real-world dynamic blurring scenarios. This dataset is widely used in the field of video deblurring, aiming to improve the video quality of applications such as autonomous driving, facial recognition and security monitoring.




