Update app.py

app.py

@@ -1,25 +1,30 @@
 from PIL import Image
 from transformers import CLIPProcessor, CLIPModel
 import gradio as gr
+import torchvision.transforms as transforms
 
 # Initialize CLIP model and processor
 processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
 model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
 
 def image_similarity(image: Image.Image, positive_prompt: str, negative_prompts: str):
-    
+    # Convert the PIL Image to a tensor and preprocess
+    transform = transforms.Compose([
+        transforms.Resize((224, 224)),
+        transforms.ToTensor(),
+        transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
+    ])
+    image_tensor = transform(image).unsqueeze(0)  # Add batch dimension
+
     # Split the negative prompts string into a list of prompts
     negative_prompts_list = negative_prompts.split(";")
-    
     # Combine positive and negative prompts into one list
     prompts = [positive_prompt.strip()] + [np.strip() for np in negative_prompts_list]
 
-    # ... rest of the code
-
-
+    # Process prompts and image tensor
     inputs = processor(
         text=prompts,
-        images=image,
+        images=image_tensor,
         return_tensors="pt",
         padding=True
     )
@@ -45,10 +50,4 @@ interface = gr.Interface(
         gr.components.Textbox(label="Probability for Positive Prompt")
     ],
     title="Engagify's Image Action Detection",
-    description="[Author: Ibrahim Hasani] This Method uses CLIP-VIT [Version: BASE-PATCH-16] to determine if an action is being performed in an image or not. (Binary Classifier). It contrasts an Action against multiple negative labels. Ensure the prompts accurately describe the desired detection.",
-    live=False,
-    theme=gr.themes.Monochrome(),
-
-)
-
-interface.launch()
+    description="[Author: Ibrahim Hasani] This Method uses CLIP-VIT [Version: BASE-PATCH-16] to determine if an action is being performed in an image or not. (Binary Classifier). It contrasts an Action against multiple negative labels. Ensure the prompts accurately describe the desired