Zocket_ImageBind.py

from imagebind import data
import torch
from imagebind.models import imagebind_model
from imagebind.models.imagebind_model import ModalityType
import gradio as gr

# command = "pip install git+https://github.com/facebookresearch/pytorchvideo.git@28fe037d212663c6a24f373b94cc5d478c8c1a1d timm==0.6.7 ftfy regex einops fvcore decord==0.6.0"
# process = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE)
# process.wait()
# print(process.returncode)  # should print 0 if installation was successful



device = "cuda:0" if torch.cuda.is_available() else "cpu"

# Instantiate model
model = imagebind_model.imagebind_huge(pretrained=True)
model.eval()
model.to(device)

text_list = ["An Advertisement(branding, text, promotions, lifestyle depiction, contextual cues, and visual composition)","Not an Advertisement"]
image_paths = []



with gr.Blocks() as demo:
    image = gr.File()
    image_paths.append(image)


    gr.Markdown(
    """
    Zocket ImageBind made AdBind
    """)


    inputs = {
        ModalityType.TEXT: data.load_and_transform_text(text_list, device),
        ModalityType.VISION: data.load_and_transform_vision_data(image_paths, device),
    }

    with torch.no_grad():
        embeddings = model(inputs)

    print(
        "Vision x Text: ",
        torch.softmax(embeddings[ModalityType.VISION] @ embeddings[ModalityType.TEXT].T, dim=-1),
    )

    out = f"""Output = {torch.softmax(embeddings[ModalityType.VISION] @ embeddings[ModalityType.TEXT].T, dim=-1)}"""
    gr.Markdown(out)



demo.launch()


# Load data