Update app.py

app.py

@@ -5,6 +5,8 @@ from transformers import BlipProcessor, BlipForConditionalGeneration  # 4. BLIP
 from PIL import Image  # 5. Pillow for image processing
 import openai  # 6. OpenAI library for DALL·E API calls
 import os  # 7. OS for environment variables
+from face_to_prompt import extract_face_prompt
+
 
 # 8. Create the FastAPI app
 app = FastAPI()
@@ -23,13 +25,18 @@ async def generate(
     style: str = Form("chibi"),         # 13. The desired style (chibi/anime/cartoon), defaults to "chibi"
 ):
     # 14. Load and convert the uploaded image to RGB
-    img_bytes = await image.read()
-    img = Image.open(io.BytesIO(img_bytes)).convert("RGB")
+    # img_bytes = await image.read()
+    # img = Image.open(io.BytesIO(img_bytes)).convert("RGB")
+
+    # # 15. Caption the image using BLIP
+    # inputs = processor(img, return_tensors="pt")
+    # out = model.generate(**inputs)
+    # caption = processor.decode(out[0], skip_special_tokens=True)
+
+    with open("temp_input.jpg", "wb") as f:
+        f.write(await image.read())
 
-    # 15. Caption the image using BLIP
-    inputs = processor(img, return_tensors="pt")
-    out = model.generate(**inputs)
-    caption = processor.decode(out[0], skip_special_tokens=True)
+    caption = extract_face_prompt("temp_input.jpg")
 
     # 16. Construct the DALL·E prompt using the style and the caption
     prompt = (