/PNN

PNN implemented in PyTorch

Primary LanguagePython

PNN

PNN pansharpenning method implemented in pytorch.

Pretrained model is provided

Based on implementation: https://github.com/xyc19970716/Deep-Learning-PanSharpening/tree/main

Paper link: https://www.mdpi.com/2072-4292/8/7/594

Dataset

The GaoFen-2 and WorldView-3 dataset download links can be found in https://github.com/liangjiandeng/PanCollection

Torch Summary

==========================================================================================
Layer (type:depth-idx)                   Output Shape              Param #
==========================================================================================
├─Conv2d: 1-1                            [-1, 64, 256, 256]        25,984
├─ReLU: 1-2                              [-1, 64, 256, 256]        --
├─Conv2d: 1-3                            [-1, 32, 256, 256]        51,232
├─ReLU: 1-4                              [-1, 32, 256, 256]        --
├─Conv2d: 1-5                            [-1, 4, 256, 256]         3,204
==========================================================================================
Total params: 80,420
Trainable params: 80,420
Non-trainable params: 0
Total mult-adds (G): 5.26
==========================================================================================
Input size (MB): 0.25
Forward/backward pass size (MB): 50.00
Params size (MB): 0.31
Estimated Total Size (MB): 50.56
==========================================================================================

Quantitative Results

GaoFen-2

alt text

WorldView-3

alt text

Qualitative Results

GaoFen-2

alt text

WorldView-3

alt text