Update app.py

app.py

@@ -116,7 +116,7 @@ pixel_mean = torch.Tensor( [123.675, 116.28, 103.53]).to(device).view(3, 1, 1)
 pixel_std = torch.Tensor([58.395, 57.12, 57.375]).to(device).view(3, 1, 1)
 normalizer = lambda x: (x - pixel_mean) / pixel_std
 inference_size = 800
-video_inference_size = 720
+video_inference_size = 640
 inference_type = 'resize_shot'  # or LSJ 
 size_divisibility = 32
 
@@ -414,11 +414,11 @@ def segment_video(video, prompt_mode, categoryname, custom_category, expressiong
     if model_selection == 'GLEE-Plus (SwinL)':
         GLEEmodel = GLEEmodel_swin
         print('use GLEE-Plus')
-        clip_length = 4 #batchsize
+        clip_length = 2 #batchsize
     else:
         GLEEmodel = GLEEmodel_r50
         print('use GLEE-Lite') 
-        clip_length = 8 #batchsize
+        clip_length = 4 #batchsize
 
     # read video and get sparse frames
     cap = cv2.VideoCapture(video)