app.py

import torch
from PIL import Image
import open_clip
import gradio as gr

def start():
    model, _, preprocess = open_clip.create_model_and_transforms(
        "ViT-B-32", pretrained="laion2b_s34b_b79k"
    )
    tokenizer = open_clip.get_tokenizer("ViT-B-32")
    return model, preprocess, tokenizer


def process(model, preprocess, tokenizer, image_path, text):
    if isinstance(image_path, str):
        image = Image.open(image_path)
        image = preprocess(image.resize((512, 512))).unsqueeze(0)
    else:
        image = preprocess(image_path.resize((512, 512))).unsqueeze(0)

    text = tokenizer([text])
    with torch.no_grad(), torch.cuda.amp.autocast():
        image_features = model.encode_image(image)
        text_features = model.encode_text(text)
        image_features /= image_features.norm(dim=-1, keepdim=True)
        text_features /= text_features.norm(dim=-1, keepdim=True)
        similarity = (image_features @ text_features.T) * 100
    return similarity


def predict(image, text):
    similarity = process(model, preprocess, tokenizer, image, text)
    return similarity.item()

gradio_app = gr.Interface(
    predict,
    inputs=[
        gr.Image(label="Select the picture", type="pil"),
        gr.Textbox(label="Enter the text"),
    ],
    outputs=gr.Textbox(label="Similarity"),
    title="You draw&AI guess",
)

if __name__ == "__main__":
    print("Start loading model...")
    model, preprocess, tokenizer = start()
    # If you want to run it locally, you can use the following code :(
    gradio_app.launch()
    # If you want to share it online, you can use the following code :)
    # gradio_app.launch(share=True)