DEM-SuperRes-Model / README.md
zcash's picture
Create README.md
d7e20bb verified
|
raw
history blame
693 Bytes
metadata
license: mit
tags:
  - pytorch
  - pytorch-lightning
  - dem
  - super-resolution
  - remote-sensing

DEM Super-Resolution Model

This is a trained PyTorch Lightning model for super-resolving 30m SRTM DEMs to 10m resolution using Sentinel-2 imagery.

Model Details

  • Architecture: U-Net with ResNet34 encoder
  • Input Channels: 7 (SRTM, RGB, NIR, NDVI, mask)
  • Output: Residual corrections for DEM super-resolution
  • Training Data: McKinley Mine LiDAR DEM
  • Inference: Applied to Marrakech, Morocco

Usage

import torch
from your_model_loading_code import DeepDEMRefinement

model = DeepDEMRefinement.load_from_checkpoint("deepdem_model.ckpt")
# Use as needed