Useful for urban planning, map updating, and infrastructure analysis. Employs image processing and machine learning to detect and map roads from high-resolution satellite data.
----------------------------------------------------------------
Layer (type) Output Shape Param #
================================================================
Conv2d-1 [-1, 16, 256, 256] 448
BatchNorm2d-2 [-1, 16, 256, 256] 32
Conv2d-3 [-1, 16, 256, 256] 2,320
BatchNorm2d-4 [-1, 16, 256, 256] 32
Dropout-5 [-1, 16, 256, 256] 0
MaxPool2d-6 [-1, 16, 128, 128] 0
Conv2d-7 [-1, 32, 128, 128] 4,640
BatchNorm2d-8 [-1, 32, 128, 128] 64
Conv2d-9 [-1, 32, 128, 128] 9,248
BatchNorm2d-10 [-1, 32, 128, 128] 64
Dropout-11 [-1, 32, 128, 128] 0
MaxPool2d-12 [-1, 32, 64, 64] 0
Conv2d-13 [-1, 48, 64, 64] 13,872
BatchNorm2d-14 [-1, 48, 64, 64] 96
Conv2d-15 [-1, 48, 64, 64] 20,784
BatchNorm2d-16 [-1, 48, 64, 64] 96
Dropout-17 [-1, 48, 64, 64] 0
MaxPool2d-18 [-1, 48, 32, 32] 0
Conv2d-19 [-1, 64, 32, 32] 27,712
BatchNorm2d-20 [-1, 64, 32, 32] 128
Conv2d-21 [-1, 64, 32, 32] 36,928
BatchNorm2d-22 [-1, 64, 32, 32] 128
Dropout-23 [-1, 64, 32, 32] 0
MaxPool2d-24 [-1, 64, 16, 16] 0
Conv2d-25 [-1, 128, 16, 16] 73,856
BatchNorm2d-26 [-1, 128, 16, 16] 256
Conv2d-27 [-1, 128, 16, 16] 147,584
BatchNorm2d-28 [-1, 128, 16, 16] 256
Dropout-29 [-1, 128, 16, 16] 0
ConvTranspose2d-30 [-1, 64, 32, 32] 32,832
Conv2d-31 [-1, 64, 32, 32] 73,792
BatchNorm2d-32 [-1, 64, 32, 32] 128
Dropout-33 [-1, 64, 32, 32] 0
Conv2d-34 [-1, 64, 32, 32] 36,928
BatchNorm2d-35 [-1, 64, 32, 32] 128
Dropout-36 [-1, 64, 32, 32] 0
ConvTranspose2d-37 [-1, 48, 64, 64] 12,336
Conv2d-38 [-1, 48, 64, 64] 41,520
BatchNorm2d-39 [-1, 48, 64, 64] 96
Dropout-40 [-1, 48, 64, 64] 0
Conv2d-41 [-1, 48, 64, 64] 20,784
BatchNorm2d-42 [-1, 48, 64, 64] 96
Dropout-43 [-1, 48, 64, 64] 0
ConvTranspose2d-44 [-1, 32, 128, 128] 6,176
Conv2d-45 [-1, 32, 128, 128] 18,464
BatchNorm2d-46 [-1, 32, 128, 128] 64
Dropout-47 [-1, 32, 128, 128] 0
Conv2d-48 [-1, 32, 128, 128] 9,248
BatchNorm2d-49 [-1, 32, 128, 128] 64
Dropout-50 [-1, 32, 128, 128] 0
ConvTranspose2d-51 [-1, 16, 256, 256] 2,064
Conv2d-52 [-1, 16, 256, 256] 4,624
BatchNorm2d-53 [-1, 16, 256, 256] 32
Dropout-54 [-1, 16, 256, 256] 0
Conv2d-55 [-1, 16, 256, 256] 2,320
BatchNorm2d-56 [-1, 16, 256, 256] 32
Dropout-57 [-1, 16, 256, 256] 0
Conv2d-58 [-1, 1, 256, 256] 17
================================================================
Total params: 600,289
Trainable params: 600,289
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.75
Forward/backward pass size (MB): 173.25
Params size (MB): 2.29
Estimated Total Size (MB): 176.29
----------------------------------------------------------------