- [MINOR] [SOURCE] [UPDATE] 1. update app.py

app.py

@@ -169,18 +169,19 @@ def inference_one_video(input_video):
 
     results = [r.mean().item() for r in evaluator(views)]
     pred_score = fuse_results(results)
-    
-    normalized_scores = [
-        normalize_score(pred_score["semantic"] , normalization_array["semantic"][0] , normalization_array["semantic"][1] ),
-        normalize_score(pred_score["technical"], normalization_array["technical"][0], normalization_array["technical"][1]),
-        normalize_score(pred_score["aesthetic"], normalization_array["aesthetic"][0], normalization_array["aesthetic"][1]),
-        normalize_score(pred_score["overall"]  , normalization_array["overall"][0]  , normalization_array["overall"][1])
-    ]
-    
+
     comparison_array["semantic"]  = pd.read_csv('./prediction_results/youtube_ugc/smos.csv')['Mos']
     comparison_array["technical"] = pd.read_csv('./prediction_results/youtube_ugc/tmos.csv')['Mos']
     comparison_array["aesthetic"] = pd.read_csv('./prediction_results/youtube_ugc/amos.csv')['Mos']
     comparison_array["overall"]   = pd.read_csv('./prediction_results/youtube_ugc/overall.csv')['Mos']
+
+    normalized_scores = [
+        normalize_score(pred_score["semantic"] , comparison_array["semantic"].min() , comparison_array["semantic"].max() ),
+        normalize_score(pred_score["technical"], comparison_array["technical"].min(), comparison_array["technical"].max()),
+        normalize_score(pred_score["aesthetic"], comparison_array["aesthetic"].min(), comparison_array["aesthetic"].max()),
+        normalize_score(pred_score["overall"]  , comparison_array["overall"].min()  , comparison_array["overall"].max()  )
+    ]
+
     
     comparisons = [
         compare_score(pred_score["semantic"], comparison_array["semantic"]),