Update app.py

app.py

@@ -24,8 +24,8 @@ os.makedirs(LR_path)
 os.makedirs(Ref_path)
 os.makedirs(Ref_path_T)
 os.makedirs('result')
-
-def resize(max_side,img):
+max_side = 512
+def resize(img):
     #basewidth = max_side
     #wpercent = (basewidth/float(img.size[0]))
     #hsize = int((float(img.size[1])*float(wpercent)))
@@ -41,8 +41,8 @@ def resize(max_side,img):
     return img
   
 def inference(LR, Ref):
-    LR = resize(256, LR)
-    Ref = resize(256, Ref)
+    LR = resize(LR)
+    Ref = resize(Ref)
 
     LR.save(os.path.join(LR_path, '0000.png'))
     Ref.save(os.path.join(Ref_path, '0000.png'))
@@ -68,8 +68,8 @@ description="Demo application for Reference-based Video Super-Resolution (RefVSR
 
 article = "<p style='text-align: center'><b>To check the full capability of the module, we recommend to clone Github repository and run RefVSR models on videos using GPUs.</b></p><p style='text-align: center'>This demo runs on CPUs and only supports RefVSR for a single LR and Ref frame due to computational complexity. Hence, the model will not take advantage of temporal LR and Ref frames.</p><p style='text-align: center'>The model is the small-sized model trained with the proposed two-stage training strategy.</p><p style='text-align: center'>The frames will be resized so that the length of a longer side of the frames doesn't exceed 256 pixels.</p><p style='text-align: center'><a href='https://junyonglee.me/projects/RefVSR' target='_blank'>Project</a> | <a href='https://arxiv.org/abs/2203.14537' target='_blank'>arXiv</a> | <a href='https://github.com/codeslake/RefVSR' target='_blank'>Github</a></p>"
 
-LR = resize(256, Image.open('LR.png'))
-Ref = resize(256, Image.open('Ref.png'))
+LR = resize(Image.open('LR.png'))
+Ref = resize(Image.open('Ref.png'))
 LR.save('LR.png')
 Ref.save('Ref.png')