Spaces:
Running
Running
| import streamlit as st | |
| import base64 | |
| import os | |
| import requests | |
| from PIL import Image | |
| from io import BytesIO | |
| # Function to compress and resize the image before base64 encoding | |
| def compress_and_resize_image(image, max_size=(1024, 1024), quality=85): | |
| img = Image.open(image) | |
| img.thumbnail(max_size) # Resize image while maintaining aspect ratio | |
| with BytesIO() as byte_io: | |
| img.save(byte_io, format="JPEG", quality=quality) # Save with reduced quality | |
| byte_io.seek(0) | |
| return byte_io | |
| # Function to convert uploaded image to base64 | |
| def convert_image_to_base64(image): | |
| compressed_image = compress_and_resize_image(image) | |
| image_bytes = compressed_image.read() | |
| encoded_image = base64.b64encode(image_bytes).decode("utf-8") | |
| return encoded_image | |
| # Function to generate caption using Nebius API | |
| def generate_caption(encoded_image): | |
| API_URL = "https://api.studio.nebius.ai/v1/chat/completions" | |
| API_KEY = os.environ.get("NEBIUS_API_KEY") | |
| headers = { | |
| "Authorization": f"Bearer {API_KEY}", | |
| "Content-Type": "application/json" | |
| } | |
| payload = { | |
| "model": "Qwen/Qwen2-VL-72B-Instruct", | |
| "messages": [ | |
| { | |
| "role": "system", | |
| "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 caption for this image" | |
| }, | |
| { | |
| "role": "user", | |
| "content": f"data:image/png;base64,{encoded_image}" # This is where the image is passed as base64 directly | |
| } | |
| ], | |
| "temperature": 0 | |
| } | |
| # Send request to Nebius API | |
| response = requests.post(API_URL, headers=headers, json=payload) | |
| if response.status_code == 200: | |
| result = response.json() | |
| caption = result.get("choices", [{}])[0].get("message", {}).get("content", "No caption generated.") | |
| return caption | |
| else: | |
| st.error(f"API Error {response.status_code}: {response.text}") | |
| return None | |
| # Streamlit app layout | |
| def main(): | |
| 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"]) | |
| if uploaded_file: | |
| # Display the uploaded image | |
| st.image(uploaded_file, caption="Uploaded Image", use_container_width=True) | |
| 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: 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") | |
| st.success("Caption generated successfully!") | |
| if __name__ == "__main__": | |
| main() | |