fix : confidence matrix

image_ref/main.py

@@ -146,6 +146,7 @@ def make_prediction_duo(model, data, f_name, f_name2):
     confidence_pred_list = [[] for i in range(n_class)]
     y_pred = []
     y_true = []
+    soft_max = nn.Softmax(dim=1)
     # iterate over test data
     for imaer,imana,img_ref, label in data:
         imaer = imaer.transpose(0,1)
@@ -163,7 +164,8 @@ def make_prediction_duo(model, data, f_name, f_name2):
             img_ref = img_ref.cuda()
             label = label.cuda()
         output = model(imaer,imana,img_ref)
-        confidence_pred_list[specie].append(output[:,0].data.cpu().numpy())
+        confidence = soft_max(output)
+        confidence_pred_list[specie].append(confidence[:,0].data.cpu().numpy())
         #Mono class output (only most postive paire)
         output = torch.argmax(output[:,0])
         label = torch.argmin(label)