RPVP-Synthetic, RPVP-Real
收藏资源简介:
RPVP-Synthetic和RPVP-Real是由宾夕法尼亚州立大学的研究团队创建的两个用于消失点检测的数据集。RPVP-Synthetic包含3200张使用Blender生成的合成图像,每张图像都带有地面实况消失点和相机参数。RPVP-Real则包含1400张真实世界图像,这些图像中的消失点由人工标注。数据集旨在解决计算机视觉中的消失点检测问题,特别是在缺乏显式直线的情况下。通过结合显式和隐式线条,该数据集为消失点检测提供了新的基准,适用于3D重建、相机校准和自动驾驶等领域。
RPVP-Synthetic and RPVP-Real are two datasets for vanishing point detection developed by a research team from Pennsylvania State University. RPVP-Synthetic contains 3200 synthetic images generated using Blender, each paired with ground-truth vanishing points and camera parameters. RPVP-Real includes 1400 real-world images, whose vanishing points were manually annotated. These datasets aim to address the vanishing point detection problem in computer vision, particularly in scenarios where explicit straight lines are scarce. By integrating explicit and implicit lines, they provide a novel benchmark for vanishing point detection, applicable to fields such as 3D reconstruction, camera calibration, and autonomous driving.

- 1Recurrence-based Vanishing Point Detection宾夕法尼亚州立大学电子工程与计算机科学学院 · 2025年



