akuratikaustiki/hackathon4a

1import math2import torch3import torchvision4import torch.nn as nn5import torch.nn.functional as F6from torchvision import transforms7# Add more imports if required8 9# Sample Transformation function10# YOUR CODE HERE for changing the Transformation values.11trnscm = transforms.Compose([transforms.Resize((100,100)), transforms.ToTensor()])12 13##Example Network14class Siamese(torch.nn.Module):15    def __init__(self):16        super(Siamese, self).__init__()17        #YOUR CODE HERE18        self.cnn1 = nn.Sequential(19            nn.ReflectionPad2d(1),       #Pads the input tensor using the reflection of the input boundary, it similar to the padding.20            nn.Conv2d(1, 4, kernel_size=3),21            nn.ReLU(inplace=True),22            nn.BatchNorm2d(4),23 24            nn.ReflectionPad2d(1),25            nn.Conv2d(4, 8, kernel_size=3),26            nn.ReLU(inplace=True),27            nn.BatchNorm2d(8),28 29 30            nn.ReflectionPad2d(1),31            nn.Conv2d(8, 8, kernel_size=3),32            nn.ReLU(inplace=True),33            nn.BatchNorm2d(8),34        )35 36        self.fc1 = nn.Sequential(37            nn.Linear(8*100*100, 500),38            nn.ReLU(inplace=True),39 40            nn.Linear(500, 500),41            nn.ReLU(inplace=True),42 43            nn.Linear(500, 10))44        45    # forward_once is for one image. This can be used while classifying the face images46    def forward_once(self, x):47        output = self.cnn1(x)48        output = output.view(output.size()[0], -1)49        output = self.fc1(output)50        return output51 52    def forward(self, input1, input2):53        output1 = self.forward_once(input1)54        output2 = self.forward_once(input2)55        return output1, output256        57##########################################################################################################58## Sample classification network (Specify if you are using a pytorch classifier during the training)    ##59## classifier = nn.Sequential(nn.Linear(64, 64), nn.BatchNorm1d(64), nn.ReLU(), nn.Linear...)           ##60##########################################################################################################61 62# YOUR CODE HERE for pytorch classifier63classifier=nn.Sequential(nn.Linear(8*100*100, 500),nn.ReLU(inplace=True),nn.Linear(500, 500),nn.ReLU(inplace=True), nn.Linear(500, 10))64# Definition of classes as dictionary65person_labels=["Aparna","Kaustiki","Gouthami","Venkatesh"]