app.py

import gradio as gr
import spaces
from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor
from qwen_vl_utils import process_vision_info
import torch
import base64
from PIL import Image
from io import BytesIO


models = {
    "Qwen/Qwen2-VL-7B-Instruct": Qwen2VLForConditionalGeneration.from_pretrained("Qwen/Qwen2-VL-7B-Instruct") #, torch_dtype="auto", device_map="auto")
}

processors = {
    "Qwen/Qwen2-VL-7B-Instruct": AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct")
}

DESCRIPTION = "# Qwen2-VL Object Localization Demo"


def image_to_base64(image):
    buffered = BytesIO()
    image.save(buffered, format="PNG")  # Save the image in memory as PNG
    img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")  # Encode image to base64
    return img_str


@spaces.GPU
def run_example(image, text_input, model_id="Qwen/Qwen2-VL-7B-Instruct"):
    model = models[model_id].eval().cuda()
    processor = processors[model_id]

    messages = [
        {
            "role": "user",
            "content": [
                {"type": "image", "image": f"data:image;base64,{image_to_base64(image)}"},
                {"type": "text", "text": f"Give a bounding box for {text_input}"},
            ],
        }
    ]

    text = processor.apply_chat_template(
        messages, tokenize=False, add_generation_prompt=True
    )
    image_inputs, video_inputs = process_vision_info(messages)
    inputs = processor(
        text=[text],
        images=image_inputs,
        videos=video_inputs,
        padding=True,
        return_tensors="pt",
    )
    inputs = inputs.to("cuda")

    generated_ids = model.generate(**inputs, max_new_tokens=128)
    generated_ids_trimmed = [
        out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
    ]
    output_text = processor.batch_decode(
        generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
    )
    return output_text

css = """
  #output {
    height: 500px; 
    overflow: auto; 
    border: 1px solid #ccc; 
  }
"""

with gr.Blocks(css=css) as demo:
    gr.Markdown(DESCRIPTION)
    with gr.Tab(label="Qwen2-VL Input"):
        with gr.Row():
            with gr.Column():
                input_img = gr.Image(label="Input Picture", type="pil")
                model_selector = gr.Dropdown(choices=list(models.keys()), label="Model", value="Qwen/Qwen2-VL-7B-Instruct")
                text_input = gr.Textbox(label="Description of Localization Target")
                submit_btn = gr.Button(value="Submit")
            with gr.Column():
                output_text = gr.Textbox(label="Output Text")

        submit_btn.click(run_example, [input_img, text_input, model_selector], [output_text])

demo.launch(debug=True)