Update app.py

app.py

@@ -2,7 +2,7 @@ import gradio as gr
 import torch
 import cv2
 from lavis.models import load_model_and_preprocess
-
+from PIL import Image
 
 # Load the Blip-Caption model
 
@@ -10,10 +10,10 @@ model, vis_processors, _ = load_model_and_preprocess(name="blip_caption", model_
 
 # Define the input and output functions for Gradio
 def generate_caption(image_file):
-        
+        image = Image.fromarray(image_file).convert('RGB')
 
         # Preprocess the image using the Blip-Caption model's visual processors
-        image = vis_processors["eval"](image_file).unsqueeze(0)
+        image = vis_processors["eval"](image).unsqueeze(0)
 
         # Generate captions using the Blip-Caption model
         captions = model.generate({"image": image}, use_nucleus_sampling=True, num_captions=5)