Upload 13 files

custom_resnet.py

@@ -0,0 +1,87 @@
+# model.py file 
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+class BasicBlock(nn.Module):
+
+    def __init__(self, in_channels, out_channels, stride=1):
+        super(BasicBlock, self).__init__()
+
+        self.conv1 = nn.Conv2d(in_channels, out_channels, kernel_size=3,
+                               stride=stride, padding=1, bias=False)
+        self.bn1 = nn.BatchNorm2d(out_channels)
+
+        self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3,
+                               stride=1, padding=1, bias=False)
+        self.bn2 = nn.BatchNorm2d(out_channels)
+
+        self.relu = nn.ReLU()
+
+    def forward(self, x):
+        x = self.relu(self.bn1(self.conv1(x)))
+        x = self.relu(self.bn2(self.conv2(x)))
+        return x
+
+class ResNet(nn.Module):
+    def __init__(self, block, num_classes=10):
+        super(ResNet, self).__init__()
+
+        self.preparation = nn.Sequential(
+                nn.Conv2d(3, 64, kernel_size=3, stride=1, padding=1, bias=False),
+                nn.BatchNorm2d(64),
+                nn.ReLU()
+                )
+
+        self.layer1 = nn.Sequential(
+                nn.Conv2d(64, 128, kernel_size=3, stride=1, padding=1, bias=False),
+                nn.MaxPool2d(2, 2),
+                nn.BatchNorm2d(128),
+                nn.ReLU()
+                )
+
+        self.residual1 = block(128, 128, 1)
+
+        self.layer2 = nn.Sequential(
+                nn.Conv2d(128, 256, kernel_size=3, stride=1, padding=1, bias=False),
+                nn.MaxPool2d(2, 2),
+                nn.BatchNorm2d(256),
+                nn.ReLU()
+                )
+
+        self.layer3 = nn.Sequential(
+                nn.Conv2d(256, 512, kernel_size=3, stride=1, padding=1, bias=False),
+                nn.MaxPool2d(2, 2),
+                nn.BatchNorm2d(512),
+                nn.ReLU()
+                )
+
+        self.residual3 = block(512, 512, 1)
+
+        self.maxpool2d = nn.MaxPool2d(4, 4)
+        self.fc = nn.Linear(512, num_classes)
+
+    def forward(self, x):
+
+        x = self.preparation(x)
+
+        x = self.layer1(x)
+        res1 = self.residual1(x)
+        x = x + res1
+
+        x = self.layer2(x)
+
+        x = self.layer3(x)
+        res3 = self.residual3(x)
+        x = x + res3
+
+        x = self.maxpool2d(x)
+        x = x.view(x.size(0), -1)
+
+        x = self.fc(x)
+
+        return x
+
+def Custom_ResNet():
+    return ResNet(BasicBlock, num_classes=10)

results/custom_resnet_trained.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0950f4cb97b78183165e325857b6c057eddd54bc82df71831879611de4f3b42
+size 26326759

utils.py

@@ -0,0 +1,53 @@
+# utils file
+
+import matplotlib.pyplot as plt
+import torch
+from torchsummary import summary
+from torchvision import transforms
+import torchvision
+import numpy as np 
+import cv2
+
+cv2.setNumThreads(0)
+cv2.ocl.setUseOpenCL(False)
+
+class Cifar10SearchDataset(torchvision.datasets.CIFAR10):
+
+    def __init__(self, root="./data", train=True, download=True, transform=None):
+      super().__init__(root=root, train=train, download=download, transform=transform)
+
+    def __getitem__(self, index):
+      image, label = self.data[index], self.targets[index]
+      
+      if self.transform is not None:
+        transformed = self.transform(image=image)
+        image = transformed["image"]
+
+        return image, label
+        
+import albumentations as A
+from albumentations.pytorch.transforms import ToTensorV2
+
+import numpy as np 
+import cv2
+
+cv2.setNumThreads(0)
+cv2.ocl.setUseOpenCL(False)
+
+def augmentation_custom_resnet(data, mu=(0.49139968, 0.48215827, 0.44653124), sigma=(0.24703233, 0.24348505, 0.26158768), pad=4):
+
+  if data == 'Train':
+    transform = A.Compose([A.PadIfNeeded(min_height=32+pad,
+                                        min_width=32+pad,
+                                        border_mode=cv2.BORDER_CONSTANT,
+                                        value=np.mean(mu)),
+                            A.RandomCrop(32, 32),
+                            A.HorizontalFlip(p=0.5),
+                            A.Cutout(max_h_size=8, max_w_size=8),
+                            A.Normalize(mean=mu, std=sigma),
+                            ToTensorV2()])
+  else:
+    transform = A.Compose([A.Normalize(mean=mu, std=sigma),
+                           ToTensorV2()])
+
+  return transform