Mars DTM Estimation

Context

This dataset is useful for doing research in the field of mars surface monocular depth estimation and it's part of the paper "An Adversarial Generative Network Designed for High-Resolution Monocular Depth Estimation from 2D HiRISE Images of Mars".
All the sources come from UAHiRISE.

Content

The dataset is composed of 250k patches where each patch is a 3-channels 512 x 512 raster. The first two channels are respectively left and right images of the stereo pair while the third channel is the DTM.
Because DTMs are saved with absolute values you have to preprocess in case you want to predict relative values.
In my project I use a monocular depth estimation model, thus, in the dataloader I converted the absolute heights to relative depths.
The Dataset size is 800 GB.

Setup

Check code section: in the notebook, there's a setup script that downloads all the files from uahirise website and then splits them into patches. There are also the Dataset class and the Dataloader for PyTorch.

Acknowledgements

Thanks UAHiRISE for sharing these beautiful Mars images.