Hey community,
I have downloaded a resnet50 pretrained model (i tried different other architectures) for image classification, and i have changed the last layer(classifier) so it can classify my 5 image classes, but it suffers from overfitting and even though i tried a simple architecture and added dropout layer, it still not able to generalize on test data, it gives very low accuracy on the test data(around 10%).
so help please 
here is the code:
`data_transforms = {
‘train’: transforms.Compose([
transforms.RandomResizedCrop(size=500, scale=(0.8, 1.0)),
transforms.RandomRotation(degrees=15),
transforms.ColorJitter(),
transforms.RandomHorizontalFlip(),
transforms.CenterCrop(size=500), # Image net standards
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406],
[0.229, 0.224, 0.225])
]),
‘test’: transforms.Compose([
transforms.Resize(size=500),
transforms.CenterCrop(size=500),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]),
}
data_dir = ‘cassava-disease’
image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x),
data_transforms[x])
for x in [‘train’, ‘test’]}
dataloaders = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size=32,
shuffle=True, num_workers=0)
for x in [‘train’, ‘test’]}
dataset_sizes = {x: len(image_datasets[x]) for x in [‘train’, ‘test’]}
class_names = image_datasets[‘train’].classes
device = torch.device(“cuda:0” if torch.cuda.is_available() else “cpu”)
def train_model(model, criterion, optimizer, scheduler, num_epochs=25):
since = time.time()
best_model_wts = copy.deepcopy(model.state_dict())
best_acc = 0.0
for epoch in range(num_epochs):
print('Epoch {}/{}'.format(epoch+1, num_epochs ))
print('-' * 10)
# Each epoch has a training and validation phase
for phase in ['train', 'test']:
if phase == 'train':
model.train() # Set model to training mode
else:
model.eval() # Set model to evaluate mode
running_loss = 0.0
running_corrects = 0
# Iterate over data.
for inputs, labels in dataloaders[phase]:
inputs = inputs.to(device)
labels = labels.to(device)
# zero the parameter gradients
optimizer.zero_grad()
# forward
# track history if only in train
with torch.set_grad_enabled(phase == 'train'):
outputs = model(inputs)
_, preds = torch.max(outputs, 1)
loss = criterion(outputs, labels)
# backward + optimize only if in training phase
if phase == 'train':
loss.backward()
optimizer.step()
# statistics
running_loss += loss.item() * inputs.size(0)
running_corrects += torch.sum(preds == labels.data)
#if phase == 'train':
# scheduler.step()
epoch_loss = running_loss / dataset_sizes[phase]
epoch_acc = running_corrects.double() / dataset_sizes[phase]
print('{} Loss: {:.4f} Acc: {:.4f}'.format(
phase, epoch_loss, epoch_acc))
# deep copy the model
if phase == 'test' and epoch_acc > best_acc:
best_acc = epoch_acc
best_model_wts = copy.deepcopy(model.state_dict())
torch.save(model, 'model.pth')
print('***Model Saved!***')
print()
time_elapsed = time.time() - since
print('Training complete in {:.0f}m {:.0f}s'.format(
time_elapsed // 60, time_elapsed % 60))
print('Best val Acc: {:4f}'.format(best_acc))
# load best model weights
model.load_state_dict(best_model_wts)
return model
model_ft = models.resnet50(pretrained=True)
model_ft
import adabound
#model_ft = models.vgg16(pretrained=True)
#model_ft.aux_logits=False
num_ftrs = model_ft.fc.in_features
for param in model_ft.parameters():
param.requires_grad = False
Here the size of each output sample is set to 5.
model_ft.fc = nn.Sequential(nn.Linear(num_ftrs, 256),
nn.ReLU(),
nn.Dropout(0.4),
nn.Linear(256, 128),
nn.ReLU(),
nn.Dropout(0.4),
nn.Linear(128, 5),
nn.LogSoftmax(dim=1))
criterion = nn.NLLLoss()
Observe that all parameters are being optimized
optimizer_ft = adabound.AdaBound(model_ft.fc.parameters(), lr=1e-3, final_lr=0.1)
#optimizer_ft = optim.Adam(model_ft.classifier.parameters(), lr=0.003)
exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft, step_size=7, gamma=0.1)
model_ft = model_ft.to(device)
model_ft = train_model(model_ft, criterion, optimizer_ft, exp_lr_scheduler,
num_epochs=100)
`