Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import requests
|
3 |
+
from PIL import Image
|
4 |
+
import streamlit as st
|
5 |
+
import torch
|
6 |
+
from huggingface_hub import login
|
7 |
+
from transformers import AutoProcessor, AutoModelForCausalLM
|
8 |
+
from diffusers import DiffusionPipeline
|
9 |
+
|
10 |
+
# Hugging Face token setup
|
11 |
+
hf_token = os.getenv('HF_AUTH_TOKEN')
|
12 |
+
if not hf_token:
|
13 |
+
raise ValueError("Hugging Face token is not set in the environment variables.")
|
14 |
+
login(token=hf_token)
|
15 |
+
|
16 |
+
# Initialize Stable Diffusion pipeline
|
17 |
+
pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-3.5-medium")
|
18 |
+
|
19 |
+
# Initialize captioning model and processor
|
20 |
+
caption_model_name = "pretrained-caption-model" # Replace with the actual model name
|
21 |
+
processor = AutoProcessor.from_pretrained(caption_model_name)
|
22 |
+
model = AutoModelForCausalLM.from_pretrained(caption_model_name)
|
23 |
+
|
24 |
+
# Move models to GPU if available
|
25 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
26 |
+
pipe.to(device)
|
27 |
+
model.to(device)
|
28 |
+
|
29 |
+
# Streamlit UI
|
30 |
+
st.title("Image Caption and Design Generator")
|
31 |
+
st.write("Upload an image or provide an image URL to generate a caption and use it to create a similar design.")
|
32 |
+
|
33 |
+
# Image upload or URL input
|
34 |
+
img_file = st.file_uploader("Choose an image...", type=["png", "jpg", "jpeg"])
|
35 |
+
img_url = st.text_input("Or provide an image URL:")
|
36 |
+
|
37 |
+
# Process the image
|
38 |
+
raw_image = None
|
39 |
+
if img_file:
|
40 |
+
raw_image = Image.open(img_file).convert("RGB")
|
41 |
+
st.image(raw_image, caption="Uploaded Image", use_column_width=True)
|
42 |
+
elif img_url:
|
43 |
+
try:
|
44 |
+
raw_image = Image.open(requests.get(img_url, stream=True).raw).convert("RGB")
|
45 |
+
st.image(raw_image, caption="Image from URL", use_column_width=True)
|
46 |
+
except Exception as e:
|
47 |
+
st.error(f"Error loading image from URL: {e}")
|
48 |
+
|
49 |
+
# Generate caption and design
|
50 |
+
if raw_image and st.button("Generate Caption and Design"):
|
51 |
+
with st.spinner("Generating caption..."):
|
52 |
+
# Generate caption
|
53 |
+
inputs = processor(raw_image, return_tensors="pt", padding=True, truncation=True, max_length=250)
|
54 |
+
inputs = {key: val.to(device) for key, val in inputs.items()}
|
55 |
+
out = model.generate(**inputs)
|
56 |
+
caption = processor.decode(out[0], skip_special_tokens=True)
|
57 |
+
st.success("Generated Caption:")
|
58 |
+
st.write(caption)
|
59 |
+
|
60 |
+
with st.spinner("Generating similar design..."):
|
61 |
+
# Generate similar design using the caption as a prompt
|
62 |
+
generated_image = pipe(caption).images[0]
|
63 |
+
st.success("Generated Design:")
|
64 |
+
st.image(generated_image, caption="Design Generated from Caption", use_column_width=True)
|