2021 (modified: 14 Nov 2022)ICLR 2021Readers: Everyone
Abstract:Differentiable rendering has paved the way to training neural networks to perform “inverse graphics” tasks such as predicting 3D geometry from monocular photographs. To train high performing models...