CVPR2023

MVImgNet: A Large-scale Dataset of Multi-view Images

Xianggang Yu, Mutian Xu, Yidan Zhang, Haolin Liu, Chongjie Ye, Yushuang Wu, Zizheng Yan, Chenming Zhu, Zhangyang Xiong, Tianyou Liang, Guanying Chen, Shuguang Cui, Xiaoguang Han

DOI 出版方

摘要

equal technical contribution † part of project lead ‡ corresponding author 1 FNii, CUHKSZ 2 SSE, CUHKSZ gaplab.cuhk.edu.cn/projects/MVImgNet Figure 1. It is time to embrace MVImgNet! We introduce MVImgnet, a large-scale dataset of multi-view images, which is efficiently collected by shooting videos of real-world objects. It enjoys 3D-aware signals from multi-view consistency, being a soft bridge between 2D and 3D vision. Through dense reconstruction on MVImgNet, we also present a large-scale real-world 3D object point cloud dataset -MVPNet. Exterior: Examples of various multi-view images in MVImgNet (see Fig. 3 more intuitively). Interior: Instances of colorful point clouds from MVPNet (see Fig. 7 more clearly) are assembled into a stereo sign of 'MVImgNet'.