app.py

import os
from huggingface_hub import login
from transformers import BlipProcessor, BlipForConditionalGeneration
from PIL import Image
import easyocr
import gradio as gr
from diffusers import DiffusionPipeline
import torch
import spaces  # Hugging Face Spaces module

from transformers import pipeline



# Get Hugging Face Token from environment variable
hf_token = os.getenv('HF_AUTH_TOKEN')
if not hf_token:
    raise ValueError("Hugging Face token is not set in the environment variables.")
login(token=hf_token)

# Load the processor and model
processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")


pipe2= pipeline("image-to-text", model="nlpconnect/vit-gpt2-image-captioning")


# Initialize the model
pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-3.5-medium")
device = "cuda" if torch.cuda.is_available() else "cpu"
pipe.to(device)
pipe2.to(device)
model.to(device)



@spaces.GPU(duration=300)
def generate_caption_and_image(image):
    img = image.convert("RGB")
    # reader = easyocr.Reader(['en'])  
    # result = reader.readtext(img)
    import random

    # Define lists for the three variables
    fabrics = ['cotton', 'silk', 'denim', 'linen', 'polyester', 'wool', 'velvet']
    patterns = ['striped', 'floral', 'geometric', 'abstract', 'solid', 'polka dots']
    textile_designs = ['woven texture', 'embroidery', 'printed fabric', 'hand-dyed', 'quilting']
    
    # Randomly select one from each category
    selected_fabric = random.choice(fabrics)
    selected_pattern = random.choice(patterns)
    selected_textile_design = random.choice(textile_designs)
    
    

      

    
    # Generate caption
    inputs = processor(image, return_tensors="pt", padding=True, truncation=True, max_length=250)
    inputs = {key: val.to(device) for key, val in inputs.items()}
    out = model.generate(**inputs)
    caption = processor.decode(out[0], skip_special_tokens=True)
    caption2 =pipe2(img)
    prompt = f'''Create a highly realistic clothing item based on the following descriptions: The design should reflect {caption1} and {caption2}, blending both themes into a single, stylish, and modern piece of clothing. Incorporate highly realistic and high-quality textures that exude sophistication, with realistic fabric lighting and fine details. Subtly hint at {selected_fabric}, featuring a {selected_pattern} motif and a {selected_textile_design} style that harmoniously balances the essence of both captions.'''



    # Generate image based on the caption
    generated_image = pipe(prompt).images[0]

    return caption, generated_image

# Gradio UI
iface = gr.Interface(
    fn=generate_caption_and_image,
    inputs=gr.Image(type="pil", label="Upload Image"),
    outputs=[gr.Textbox(label="Generated Caption"), gr.Image(label="Generated Design")],
    live=True
)
iface.launch(share=True)