PNN pansharpenning method pytorch implementation
Based on implementation: https://github.com/xyc19970716/Deep-Learning-PanSharpening/tree/main
Paper link: https://ieeexplore.ieee.org/abstract/document/8127731
==========================================================================================
Layer (type:depth-idx) Output Shape Param #
==========================================================================================
MSDCNN_model [1, 8, 256, 256] --
├─Conv2d: 1-1 [1, 64, 256, 256] 46,720
├─ReLU: 1-2 [1, 64, 256, 256] --
├─Conv2d: 1-3 [1, 32, 256, 256] 2,080
├─ReLU: 1-4 [1, 32, 256, 256] --
├─Conv2d: 1-5 [1, 8, 256, 256] 6,408
├─Conv2d: 1-6 [1, 60, 256, 256] 26,520
├─ReLU: 1-7 [1, 60, 256, 256] --
├─Conv2d: 1-8 [1, 20, 256, 256] 10,820
├─ReLU: 1-9 [1, 20, 256, 256] --
├─Conv2d: 1-10 [1, 20, 256, 256] 30,020
├─ReLU: 1-11 [1, 20, 256, 256] --
├─Conv2d: 1-12 [1, 20, 256, 256] 58,820
├─ReLU: 1-13 [1, 20, 256, 256] --
├─Conv2d: 1-14 [1, 30, 256, 256] 16,230
├─ReLU: 1-15 [1, 30, 256, 256] --
├─Conv2d: 1-16 [1, 10, 256, 256] 2,710
├─ReLU: 1-17 [1, 10, 256, 256] --
├─Conv2d: 1-18 [1, 10, 256, 256] 7,510
├─ReLU: 1-19 [1, 10, 256, 256] --
├─Conv2d: 1-20 [1, 10, 256, 256] 14,710
├─ReLU: 1-21 [1, 10, 256, 256] --
├─Conv2d: 1-22 [1, 8, 256, 256] 6,008
==========================================================================================
Total params: 228,556
Trainable params: 228,556
Non-trainable params: 0
Total mult-adds (G): 14.98
==========================================================================================
Input size (MB): 0.39
Forward/backward pass size (MB): 153.09
Params size (MB): 0.91
Estimated Total Size (MB): 154.40
==========================================================================================
