Update app.py

app.py

@@ -5,46 +5,42 @@ import requests
 import numpy as np
 from PIL import Image
 
-catgs = [
-  "Shirts",
-  "SetShirtsPants",
-  "SetJacketsPants",
-  "Pants",
-  "Jeans",
-  "JacketsCoats",
-  "Shoes",
-  "Underpants",
-  "Socks",
-  "Hats",
-  "Wallets",
-  "Bags",
-  "Scarfs",
-  "Parasols&Umbrellas",
-  "Necklaces",
-  "Towels&Robes",
-  "WallObjects",
-  "Rugs",
-  "Glassware",
-  "Mugs&Cups",
-  "OralCare"
-]
 
 model, preprocess_train, preprocess_val = open_clip.create_model_and_transforms('hf-hub:Marqo/marqo-fashionSigLIP')
 tokenizer = open_clip.get_tokenizer('hf-hub:Marqo/marqo-fashionSigLIP')
 
-text = tokenizer(catgs)
-
-with torch.no_grad(), torch.cuda.amp.autocast():
-    text_features = model.encode_text(text)
-    text_features /= text_features.norm(dim=-1, keepdim=True)
-
-
 def predict(inp):
-  image = preprocess_val(inp).unsqueeze(0)
+    catgs = [
+      "Shirts",
+      "SetShirtsPants",
+      "SetJacketsPants",
+      "Pants",
+      "Jeans",
+      "JacketsCoats",
+      "Shoes",
+      "Underpants",
+      "Socks",
+      "Hats",
+      "Wallets",
+      "Bags",
+      "Scarfs",
+      "Parasols&Umbrellas",
+      "Necklaces",
+      "Towels&Robes",
+      "WallObjects",
+      "Rugs",
+      "Glassware",
+      "Mugs&Cups",
+      "OralCare"
+    ]
+    text = tokenizer(catgs)
+    image = preprocess_val(inp).unsqueeze(0)
 
     with torch.no_grad(), torch.cuda.amp.autocast():
         image_features = model.encode_image(image)
-        image_features /= image_features.norm(dim=-1, keepdim=True)    
+        image_features /= image_features.norm(dim=-1, keepdim=True) 
+        text_features = model.encode_text(text)
+        text_features /= text_features.norm(dim=-1, keepdim=True)
         text_probs = (100.0 * image_features @ text_features.T).softmax(dim=-1)
     
     max_prob_idx = np.argmax(text_probs)