3Ddataset, synthetic dataset, 3D vision, single-view reconstruction
Abstract
Abundance in 2D segmentation datasets has enabled the training of accurate 2Dperception models. However, collecting real-world 3D perception datasets is challenging and time-consuming, and therefore, 3D perception datasets are few and often contain limited categories of objects. In this project, we aim at collecting a large-scale synthetic dataset, the GTAMesh dataset, containing annotations for 3D geometry of objects represented as meshes. To this end, we use the game engine GTA-V for collecting the data. The mesh information of variegated objects can be extracted by hooking into the rendering pipeline. We show some application cases and develop a temporal baseline model for 3D reconstruction to demonstrate the effectiveness of our dataset.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
