Model class for "Self-supervised surround-view depth estimation with volumetric feature fusion"

Trainer class for training and evaluation

This class decodes encoded 2D features to estimate depth map. Unlike monodepth depth decoder, we decode features with corresponding level we

Surround-view fusion module that estimates a single 3D feature using surround-view images

Superclass for DGP dataset loaders of the packnet-sfm repository.

Class for depth synthesis loss calculation

Depth fusion module

Canonical motion estimation module.

Logger class to monitor training

Pytorch module for a depth network from the paper "Digging into Self-Supervised Monocular Depth Prediction"

Pytorch module for a pose network from the paper "Digging into Self-Supervised Monocular Depth Prediction"

Class for multi-camera(spatio & temporal) loss calculation

Loaders for NuScenes dataset

Class for multi-camera pose calculation

This class computes projection and reprojection function.

Class for single camera(temporal only) loss calculation

Class for rendering images from given camera parameters and pixel wise depth information

Base class loss calculation