Broekman André, Gräbe Petrus Johannes
Department of Civil Engineering, University of Pretoria, South Africa - University of Pretoria, Lynnwood Road, Hatfield, Pretoria 0002, South Africa.
Data Brief. 2020 Aug 24;32:106219. doi: 10.1016/j.dib.2020.106219. eCollection 2020 Oct.
A Perfectly Accurate, Synthetic dataset for Multi-View Stereopsis (PASMVS) is presented, consisting of 400 scenes and 18,000 model renderings together with ground truth depth maps, camera intrinsic and extrinsic parameters, and binary segmentation masks. Every scene is rendered from 45 different camera views in a circular pattern, using Blender's path-tracing rendering engine. Every scene is composed from a unique combination of two camera focal lengths, four 3D models of varying geometrical complexity, five high definition, high dynamic range (HDR) environmental textures to replicate photorealistic lighting conditions and ten materials. The material properties are primarily specular, with a selection of more diffuse materials for reference. The combination of highly specular and diffuse material properties increases the reconstruction ambiguity and complexity for MVS reconstruction algorithms and pipelines, and more recently, state-of-the-art architectures based on neural network implementations. PASMVS serves as an addition to the wide spectrum of available image datasets employed in computer vision research, improving the precision required for novel research applications.
提出了一个用于多视图立体视觉的完美精确合成数据集（PASMVS），它由400个场景和18000个模型渲染图以及地面真值深度图、相机内参和外参，还有二进制分割掩码组成。每个场景使用Blender的路径追踪渲染引擎以圆形模式从45个不同的相机视图进行渲染。每个场景由两个相机焦距、四个几何复杂度不同的3D模型、五个用于复制逼真光照条件的高清晰度、高动态范围（HDR）环境纹理以及十种材质的独特组合构成。材质属性主要是镜面反射的，还选择了一些更具漫反射的材质作为参考。高镜面反射和漫反射材质属性的组合增加了多视图立体视觉（MVS）重建算法和管道，以及最近基于神经网络实现的先进架构的重建模糊性和复杂性。PASMVS作为计算机视觉研究中广泛使用的可用图像数据集的补充，提高了新研究应用所需的精度。
