[Kaggle] CT Medical Image - (6) 결과 정리 및 시각화

 

CT Medical Image Classification

(0) 서론, 액션플랜
(1) DICOM 파일 array로 전환 및 시각화
(2) DataLoader 만들기
(3) 모델 만들기
(4) 성능 지표 만들기      
(5) Augmentation

 

model.train()
criterion = nn.BCELoss().to(device)
optimizer = torch.optim.SGD(model.parameters(), lr=0.001)
##############################################
loss_values = []
##############################################
epoch = 10

for i in range(epoch):
    avg_cost = 0.0
    for e, data in enumerate(dataloader):

        image, label = data
        image = image.reshape(2,512,512,1) #채널수가 흑백=>마지막에 1
        
        image = image.permute(0,3,1,2) #batch, channel, width, height 순으로 텐서 변경
        
        image = image.to(device)
        label = label.to(device)
        
        image.requires_grad = True
    
        optimizer.zero_grad()

        outputs = model(image)

        loss = criterion(outputs, label)
        loss.backward()
        optimizer.step()
        
        avg_cost += loss / total_batch
        
    print(f'[Epoch:{i + 1}] cost = {avg_cost}')
    ##############################################
    loss_values.append(avg_cost.item())
    ##############################################
    
print('Learning Finished!')

 

 

 

 성능은 대 참패