Update app.py

app.py

@@ -44,7 +44,16 @@ def inference_automatic(input_img, class_names):
     
     class_names = class_names.split(',')
     img = read_image(input_img, format="BGR")
-    _, visualized_output = demo.run_on_image(img, class_names)
+    if len(class_names) == 1:
+        class_names.append('others')
+    txts = [f'a photo of {cls_name}' for cls_name in class_names]
+    text = open_clip.tokenize(txts)
+
+
+    text_features = clip_model.encode_text(text.cuda())
+    text_features /= text_features.norm(dim=-1, keepdim=True)
+        
+    _, visualized_output = demo.run_on_image(img, class_names,text_features)
 
     return Image.fromarray(np.uint8(visualized_output.get_image())).convert('RGB')