I have been doing the project of Knee osteoarthritis severity grading and using Cnn ( Can't use pretrained model). I am using the dataset of kaggle. I used 3 convolutional layer with 2 pooling layer .But I am facing over fitting .Most of the data in my class 1 are predicted as 0 and 2 and so is class 4 . I implemented smote class weight used different augmentations batchnormalization dropouts ,tried different learning rates , scheduler but it doesn't improve a bit . What should I do next ? this is my model without any dropouts or normalization. class SimpleCNN(nn.Module): def __init__(self,num_classes): super(SimpleCNN,self).__init__() self.conv1 = nn.Conv2d(1,64,kernel_size=3,padding=1) self.relu1 = nn.ReLU() self.pool1 = nn.MaxPool2d(kernel_size=2,stride=2) self.conv2 = nn.Conv2d(64,64,kernel_size=3,padding=1) self.relu2 = nn.ReLU() self.pool2 = nn.MaxPool2d(kernel_size=2,stride=2) self.conv3 = nn.Conv2d(64,64,kernel_size=3,padding=1) self.relu3 = nn.ReLU() self.pool3 = nn.MaxPool2d(kernel_size=2,stride=2) self.flatten = nn.Flatten() self.fc1 = nn.Linear(40000,1024) self.relu4 = nn.ReLU() self.fc2 = nn.Linear(1024,num_classes) def forward(self,x): x = self.pool1(self.relu1(self.conv1(x))) x = self.pool2(self.relu2(self.conv2(x))) x = self.pool3(self.relu3(self.conv3(x))) # print(x.size()) x = self.flatten(x) x = self.relu4(self.fc1(x)) x = self.fc2(x) return x Continue reading...
