diff --git a/ieemoo-ai-filtervideo.py b/ieemoo-ai-filtervideo.py
index f572fdf..29acc38 100644
--- a/ieemoo-ai-filtervideo.py
+++ b/ieemoo-ai-filtervideo.py
@@ -20,4 +20,4 @@ def filtervideo():
     return result
             
 if __name__ == '__main__':
-    app.run()
+    app.run(host='0.0.0.0', port=8085)
diff --git a/segpredict.py b/segpredict.py
index 8313051..4d3e869 100644
--- a/segpredict.py
+++ b/segpredict.py
@@ -1,6 +1,7 @@
 import os
 os.environ['CUDA_VISIBLE_DEVICES'] = '0'
 
+import time
 import numpy as np
 import cv2
 import matplotlib.pyplot as plt
@@ -127,8 +128,8 @@ if __name__ == '__main__':
     print(type(img_test))
     print('>>>>>>shape {}'.format(img_test.shape))
 
-    #ENCODER = 'resnet18'
-    ENCODER = 'mobilenet_v2'
+    ENCODER = 'resnet18'
+    #ENCODER = 'mobilenet_v2'
     ENCODER_WEIGHTS = 'imagenet'
     CLASSES = ['front']
     ACTIVATION = 'sigmoid' # could be None for logits or 'softmax2d' for multiclass segmentation
@@ -160,11 +161,12 @@ if __name__ == '__main__':
         image = predict_dataset[i]
 
         # 通过图像分割得到的0-1图像pr_mask
+        T1 = time.time()
         x_tensor = torch.from_numpy(image).to(DEVICE).unsqueeze(0)
         pr_mask = best_model.predict(x_tensor)
+        T2 = time.time()
+        print('>>>>>> {}'.format(T2-T1))
         pr_mask = (pr_mask.squeeze().cpu().numpy().round())
-        print('>>>>>>> pr_mask{}'.format(pr_mask.shape))
-        print('>>>>>>{} {}'.format(height, weight))
 
         # 恢复图片原来的分辨率
         #image_vis = cv2.resize(image_vis, (weight, height))
diff --git a/segtrain.py b/segtrain.py
index 4be585a..5c72e1f 100644
--- a/segtrain.py
+++ b/segtrain.py
@@ -107,7 +107,7 @@ def get_training_augmentation():
 
 def get_validation_augmentation():
     test_transform = [
-        albu.PadIfNeeded(384, 480)
+        albu.PadIfNeeded(512, 512)
     ]
     return albu.Compose(test_transform)
 
@@ -142,8 +142,8 @@ if __name__ == '__main__':
     y_valid_dir = os.path.join(DATA_DIR, 'valannot')
 
     #ENCODER = 'se_resnext50_32x4d'
-    #ENCODER = 'resnet18'
-    ENCODER = 'mobilenet_v2'
+    ENCODER = 'resnet18'
+    #ENCODER = 'mobilenet_v2'
     ENCODER_WEIGHTS = 'imagenet'
     CLASSES = ['front', 'background']
     ACTIVATION = 'sigmoid' # could be None for logits or 'softmax2d' for multiclass segmentation