Models: UTKFace¶
-
class
model.
AgeModel
¶ - Description:
- The model is a CNN with 6 convolutional layers and 1 linear layer. The images has been transformed to gray before feeding them to the network,
- Model:
Conv 1: - Conv layer
in_channels = 1, out_channels =10, kernel_size= 3, Stride= 1, Padding= 1
- BatchNorm2d
num_features= 10
- ReLU
a non linear function.
- MaxPool2d
kernel_size=2
Conv 2: - Conv layer
in_channels = 10, out_channels =20, kernel_size= 3, Stride= 1, Padding= 1
- BatchNorm2d
num_features= 20
- ReLU
a non linear function.
- MaxPool2d
kernel_size=2, stride =2
Conv 3: - Conv layer
in_channels = 20, out_channels =32, kernel_size= 3, Stride= 1, Padding= 1
- BatchNorm2d
num_features= 32
- ReLU
a non linear function.
- MaxPool2d
kernel_size=2, stride =2
Conv 4: - Conv layer
in_channels = 32, out_channels =64, kernel_size= 3, Stride= 1, Padding= 1
- BatchNorm2d
num_features= 64
- ReLU
a non linear function.
- MaxPool2d
kernel_size=2, stride =2
Conv 5: - Conv layer
in_channels = 64, out_channels =128, kernel_size= 3, Stride= 1, Padding= 1
- BatchNorm2d
num_features= 128
- ReLU
a non linear function.
- MaxPool2d
kernel_size=2, stride =2
Conv 6: - Conv layer
in_channels = 128, out_channels =256, kernel_size= 3, Stride= 1, Padding= 1
- BatchNorm2d
num_features= 256
- ReLU
a non linear function.
- MaxPool2d
kernel_size=2, stride =2
Linear: in_features = 256*3*3 , out_features = 1
-
forward
(x)¶ - Description:
- Predict an age using an image.
- Return:
- Batch of labels (ages)
- Return Type:
- Tensor
- Args:
x (tensor): Batch of images.