BiasBench
收藏资源简介:
BiasBench是一个用于调整事件相机偏置的复现性基准数据集,由德国蒂宾根大学认知系统组创建。该数据集包含多个场景,每个场景的偏置设置都以网格状模式采样。数据集包含三种不同的场景,每种场景都有一个质量指标,用于评估下游应用。此外,数据集还提供了一个基于强化学习的在线偏置调整方法。BiasBench旨在帮助研究人员开发自动调整事件相机偏置的算法,并促进该领域的发展。
BiasBench is a reproducible benchmark dataset for adjusting the bias of event cameras, developed by the Cognitive Systems Group at the University of Tübingen, Germany. This dataset encompasses multiple scenarios, where the bias settings for each scenario are sampled in a grid pattern. The dataset includes three distinct scenarios, each paired with a dedicated quality metric for evaluating downstream applications. In addition, the dataset also provides an online bias adjustment method based on reinforcement learning. BiasBench is designed to assist researchers in developing algorithms for automatically adjusting event camera biases and to promote the advancement of this research field.




