Update app.py

app.py

@@ -1,19 +1,17 @@
 import streamlit as st
 import base64
-import requests
 import os
-import random
-import time
+import requests
 
-# Function to convert image to base64
+# Function to convert uploaded image to base64
 def convert_image_to_base64(image):
     image_bytes = image.read()
     encoded_image = base64.b64encode(image_bytes).decode("utf-8")
     return encoded_image
 
-# Function to generate a caption using Nebius API
+# Function to generate caption using Nebius API
 def generate_caption(encoded_image):
-    API_URL = "https://api.studio.nebius.ai/v1/"
+    API_URL = "https://api.studio.nebius.ai/v1/chat/completions"
     API_KEY = os.environ.get("NEBIUS_API_KEY")
 
     headers = {
@@ -21,31 +19,26 @@ def generate_caption(encoded_image):
         "Content-Type": "application/json"
     }
 
-    # Add a random string to the URL to prevent caching issues
-    random_string = str(random.randint(100000, 999999))
-
     payload = {
         "model": "Qwen/Qwen2-VL-72B-Instruct",
         "messages": [
             {
                 "role": "system",
-                "content": """describe this image in great detail"""
+                "content": """You are an image to prompt converter. Your work is to observe each and every detail of the image and craft a detailed prompt under 75 words in this format: [image content/subject, description of action, state, and mood], [art form, style], [artist/photographer reference if needed], [additional settings such as camera and lens settings, lighting, colors, effects, texture, background, rendering]."""
             },
             {
                 "role": "user",
-                "content": "write a detailed caption for this image"
+                "content": "Write a caption for this image"
+            },
+            {
+                "role": "user",
+                "content": f"data:image/png;base64,{encoded_image}"  # This is where the image is passed as base64 directly
             }
         ],
-        "image": {
-            "type": "image_url",
-            "image_url": {
-                "url": f"data:image/png;base64,{encoded_image}?{random_string}"  # Added random query to avoid cache
-            }
-        },
-        "temperature": 1
+        "temperature": 0
     }
 
-    # Send request to API
+    # Send request to Nebius API
     response = requests.post(API_URL, headers=headers, json=payload)
 
     if response.status_code == 200:
@@ -56,12 +49,10 @@ def generate_caption(encoded_image):
         st.error(f"API Error {response.status_code}: {response.text}")
         return None
 
-# Streamlit app
+# Streamlit app layout
 def main():
-    st.set_page_config(page_title="Image to Caption Converter", layout="centered", initial_sidebar_state="collapsed")
-
-    st.title("🖼️ Image to Caption Converter")
-    st.markdown("Upload an image and let the AI generate a detailed caption for it.")
+    st.set_page_config(page_title="Image Caption Generator", layout="centered", initial_sidebar_state="collapsed")
+    st.title("🖼️ Image to Caption Generator")
 
     uploaded_file = st.file_uploader("Upload an image", type=["jpg", "jpeg", "png"])
 
@@ -69,22 +60,21 @@ def main():
         # Display the uploaded image
         st.image(uploaded_file, caption="Uploaded Image", use_container_width=True)
 
-        # Convert image to base64 and get caption
         if st.button("Generate Caption"):
+            # Convert the uploaded image to base64
             with st.spinner("Generating caption..."):
                 encoded_image = convert_image_to_base64(uploaded_file)
 
-                # Debugging: print out the base64 string length to verify if it's changing
+                # Debugging: Ensure the encoded image is valid and not too large
                 st.write(f"Encoded image length: {len(encoded_image)} characters")
 
+                # Get the generated caption from the API
                 caption = generate_caption(encoded_image)
 
                 if caption:
                     st.subheader("Generated Caption:")
                     st.text_area("", caption, height=100, key="caption_area")
-
-                    # Copy button (if you want to implement this, use st.session_state)
-                    st.button("Copy to Clipboard", on_click=lambda: st.session_state.update({"caption": caption}))
+                    st.success("Caption generated successfully!")
 
 if __name__ == "__main__":
     main()