change back to streamlit

README.md

@@ -3,8 +3,8 @@ title: Diary-AI
 emoji: 😃
 colorFrom: red
 colorTo: blue
-sdk: gradio
-sdk_version: "4.44.0"
+sdk: streamlit
+sdk_version: "1.38.0"
 app_file: app.py
 pinned: false
 ---

app.py

@@ -1,4 +1,4 @@
-import gradio as gr
+import streamlit as st
 import openai
 import json
 from PIL import Image
@@ -7,19 +7,30 @@ from baseline_utils import detect_text_in_image, summarize_diary_text, analyze_w
 import glob
 import os
 
-# Load secrets from Hugging Face Spaces environment
-openai_api_key = os.getenv("OPENAI_API_KEY")
-google_service_account_info = json.loads(os.getenv("GOOGLE_SERVICE_ACCOUNT"))
-gemini_api_key = os.getenv("GEMINI_API_KEY")
+# Load secrets
+openai_api_key = st.secrets["openai_api_key"]
+google_service_account_info = json.loads(st.secrets["google_service_account"])
+gemini_api_key = st.secrets["gemini_api_key"]
 
+# Initialize OpenAI
+openai.api_key = openai_api_key
 
 # Function to get Google credentials
 def get_google_credentials():
     return service_account.Credentials.from_service_account_info(google_service_account_info)
 
+st.title('Handwritten Diary to Comic Book')
+uploaded_diary = st.file_uploader("Upload your handwritten diary image", type=["png", "jpg", "jpeg"])
+uploaded_writer_image = st.file_uploader("Upload a photo of the writer", type=["png", "jpg", "jpeg"])
 
-def process_images(diary_image, writer_image):
-    # Save the file-like objects as image files
+if uploaded_diary and uploaded_writer_image:
+    st.write("Analyzing your diary and writer...")
+
+    # Read the uploaded images using file-like objects
+    diary_image = Image.open(uploaded_diary)
+    writer_image = Image.open(uploaded_writer_image)
+
+    # Save the file-like objects as image files (optional if needed)
     diary_image_path = "temp_upload_images/temp_diary_image.png"
     writer_image_path = "temp_upload_images/temp_writer_image.png"
     os.makedirs("temp_upload_images", exist_ok=True)
@@ -30,39 +41,25 @@ def process_images(diary_image, writer_image):
     google_credentials = get_google_credentials()
     detected_text = detect_text_in_image(diary_image_path, google_credentials)
     summarized_text = summarize_diary_text(detected_text, openai_api_key)
+    st.write(f"Summarized Diary Text: {summarized_text}")
 
     # Analyze the writer's image using Gemini API
     writer_summary = analyze_writer_image(writer_image_path, gemini_api_key)
+    st.write(f"Writer Description: {writer_summary}")
 
     # Generate the comic book based on the summaries
+    st.write("Generating comic book images...")
     generate_comic_book(summarized_text, writer_summary, num_pages=4)
 
+    st.write("Comic book generated successfully!")
+
     # Assuming generated images are saved as 'comic_book/page_1.png', 'comic_book/page_2.png', etc.
     image_files = sorted(glob.glob("comic_book/page_*.png"))  # Find all the generated comic book pages
 
-    return image_files
-
-
-# Define the Gradio interface
-def gradio_interface(diary_image, writer_image):
-    # Process the images and generate comic book pages
-    generated_images = process_images(diary_image, writer_image)
-
-    # Load the images and return them
-    images = [Image.open(img) for img in generated_images]
-    return images
-
-
-# Set up the Gradio interface
-interface = gr.Interface(
-    fn=gradio_interface,
-    inputs=[
-        gr.Image(label="Upload your handwritten diary image", type="pil"),
-        gr.Image(label="Upload a photo of the writer", type="pil"),
-    ],
-    outputs=gr.Gallery(label="Generated Comic Book Pages"),
-    title="Handwritten Diary to Comic Book"
-)
+    # Display images in 2 columns
+    cols = st.columns(2)  # Create two columns for the images
 
-# Launch the interface
-interface.launch()
+    for i, image_file in enumerate(image_files):
+        with cols[i % 2]:  # Alternate between the two columns
+            # Display each comic book page in the respective column
+            st.image(image_file, caption=image_file.split('/')[-1], use_column_width=True)

baseline_utils.py

@@ -79,7 +79,7 @@ def generate_comic_book(diary_text, writer_description, num_pages=4):
         "stabilityai/sdxl-turbo",
         torch_dtype=torch.float16,
         variant="fp16",
-        #cache_dir="./SDXL-Turbo"
+        cache_dir="./SDXL-Turbo"
     )
 
     # Check for available device: CUDA, MPS, or CPU

requirements.txt

@@ -5,5 +5,5 @@ google-generativeai
 diffusers
 torch
 gradio
-transformers == 4.22.0
+transformers
 accelerate