search

karfly / learnable-triangulation-pytorch

This repository is an official PyTorch implementation of the paper "Learnable Triangulation of Human Pose" (ICCV 2019, oral). Proposed method archives state-of-the-art results in multi-view 3D human pose estimation!
MIT License
1.09k stars 181 forks source link

About the dimension #140

Open CheungBH opened 3 years ago

CheungBH commented 3 years ago

Hello. Thanks for your great job. I am currently reading your paper, but I felt confusion about some details about volumetric method. You said in your paper while illustrating the cube that "We discretize the bounding box by a volumetric cube Vcoords ∈ R 64,64,64,3, filling it with the global coordinates of the center of each voxel" Could you kindly explain the physical meaning of the 4-dimension cube? I know that the (64, 64, 64) means 3 dimensions of (length, width, height). But what's the physical meaning of the forth dimension 3?

yurymalkov commented 3 years ago

The last dimension are the (x,y,z) coordinates in meters. E.g. Vcoords[0,0,0,0] corresponds to X coordinate of the zero voxel, Vcoords[0,0,0,1] corresponds to Y coordinate of the zero voxel and so on.