Added a MultiSpectral extension for VGG16 for PAD

bob/learn/pytorch/architectures/DeepMSPAD.py

+import torch
+from torch import nn
+from torchvision import models
+
+
+class DeepMSPAD(nn.Module):
+
+	""" Deep multispectral PAD algorithm
+
+	Attributes:
+		pretrained: bool
+			if set `True` loads the pretrained vgg16 model. 
+		vgg: :py:class:`torch.nn.Module`
+			The VGG16 model
+		relu: :py:class:`torch.nn.Module`
+			ReLU activation
+		enc: :py:class:`torch.nn.Module`
+			Uses the layers for feature extraction
+		linear1: :py:class:`torch.nn.Module`
+			Fully connected layer
+		linear2: :py:class:`torch.nn.Module`
+			Fully connected layer
+		dropout: :py:class:`torch.nn.Module`
+			Dropout layer
+		sigmoid: :py:class:`torch.nn.Module`
+			Sigmoid activation
+	"""
+
+	def __init__(self, pretrained=True, num_channels=4):
+
+		""" Init method
+
+		Parameters
+		----------
+		pretrained: bool
+			if set `True` loads the pretrained vgg16 model.
+		num_channels: int
+			Number of channels in the input
+
+		"""
+		super(DeepMSPAD, self).__init__()
+
+		vgg = models.vgg16(pretrained=pretrained)
+
+		features = list(vgg.features.children())
+
+		features[0]=nn.Conv2d(num_channels, 64, kernel_size=(3, 3), stride=(1, 1), padding =(1, 1))
+
+		self.enc = nn.Sequential(*features)
+
+		self.linear1=nn.Linear(25088,256)
+
+		self.relu=nn.ReLU()
+
+		self.dropout= nn.Dropout(p=0.5)
+
+		self.linear2=nn.Linear(256,1)
+
+		self.sigmoid= nn.Sigmoid()
+
+
+	def forward(self, x):
+		""" Propagate data through the network
+
+		Parameters
+		----------
+		x: :py:class:`torch.Tensor` 
+		  The data to forward through the network
+
+		Returns
+		-------
+		x: :py:class:`torch.Tensor` 
+		  The last layer of the network
+		
+		"""
+	   
+		enc = self.enc(x)
+
+		x=enc.view(-1,25088)
+
+		x=self.linear1(x)
+
+		x=self.relu(x)
+
+		x=self.dropout(x)
+
+		x=self.linear2(x)
+
+		x=self.sigmoid(x)  
+
+		return x

bob/learn/pytorch/architectures/__init__.py

@@ -6,6 +6,7 @@ from .LightCNN import LightCNN29v2
 from .MCCNN import MCCNN
 from .MCCNNv2 import MCCNNv2
 from .FASNet import FASNet
+from .DeepMSPAD import DeepMSPAD
 
 from .DCGAN import DCGAN_generator
 from .DCGAN import DCGAN_discriminator

bob/learn/pytorch/test/test.py

@@ -84,6 +84,14 @@ def test_architectures():
   output = net.forward(t)
   assert output.shape == torch.Size([1, 1])
 
+  #DeepMSPAD 
+  a = numpy.random.rand(1, 8, 224, 224).astype("float32")
+  t = torch.from_numpy(a)
+  from ..architectures import DeepMSPAD
+  net = FASNet(pretrained=False, num_channels=8)
+  output = net.forward(t)
+  assert output.shape == torch.Size([1, 1])
+
   # DCGAN
   d = numpy.random.rand(1, 3, 64, 64).astype("float32")
   t = torch.from_numpy(d)