Abstract
Appearance-based gaze estimation systems have shown great progress recently, yet the performance of these techniques depend on the datasets used for training. Most of the existing gaze estimation datasets setup in interactive settings were recorded in laboratory conditions and those recorded in the wild conditions display limited head pose and illumination variations. Further, we observed little attention so far towards precision evaluations of existing gaze estimation approaches. In this work, we present a large gaze estimation dataset, PARKS-Gaze, with wider head pose and illumination variation and with multiple samples for a single Point of Gaze (PoG). The dataset contains 974 minutes of data from 28 participants with a head pose range of 60 degrees in both yaw and pitch directions. Our within-dataset and cross-dataset evaluations and precision evaluations indicate that the proposed dataset is more challenging and enable models to generalize on unseen participants better than the existing in-the-wild datasets. The project page can be accessed here: this https URL
Abstract (translated)
appearance-based gaze estimation systems 在最近取得了巨大的进展,但这些技巧的性能取决于用于训练的 datasets。大多数现有的 gaze estimation dataset 在互动设置中设置是在实验室条件下录制的,而在野生条件下录制的dataset 显示 head pose 和照明变化非常有限。此外,我们观察到迄今为止 little 关注于 precision 评估 existing gaze estimation approaches。在本研究中,我们提出了一个大型 gaze estimation dataset,PARKS-Gaze,具有更广泛的 head pose 和照明变化,并且具有多个样本,以单点 gaze (PoG)。dataset 包含 974 分钟的数据,来自 28 名参与者,其 head pose 范围在 yaw 和 pitch 方向上为 60 度。我们的dataset 内部和跨dataset 评估以及 precision 评估表明, proposed dataset 更具挑战性,使模型能够对未观测到的参与者更普遍地泛化。项目页面可访问此处:这个 https URL。
URL
https://arxiv.org/abs/2302.02353