app.py

import gradio as gr
import numpy as np
import torch
from typing import List
from PIL import Image, ImageDraw
from transformers import OwlViTProcessor, OwlViTForObjectDetection

processor = OwlViTProcessor.from_pretrained("google/owlvit-base-patch32")
model = OwlViTForObjectDetection.from_pretrained("google/owlvit-base-patch32")


def pro_process(labelstring):
    labels = labelstring.split(",")
    labels = [i.strip() for i in labels]
    return labels


def inference(img: Image.Image, labels: List[str]) -> Image.Image:
    labels = pro_process(labels)
    print(labels)
    inputs = processor(text=labels, images=img, return_tensors="pt")
    outputs = model(**inputs)
    target_sizes = torch.Tensor([img.size[::-1]])
    results = processor.post_process_object_detection(outputs=outputs, target_sizes=target_sizes, threshold=0.1)
    i = 0
    boxes, scores, labels_index = results[i]["boxes"], results[i]["scores"], results[i]["labels"]
    draw = ImageDraw.Draw(img)
    for box, score, label_index in zip(boxes, scores, labels_index):
        box = [round(i, 2) for i in box.tolist()]
        xmin, ymin, xmax, ymax = box
        draw.rectangle((xmin, ymin, xmax, ymax), outline="red", width=1)
        draw.text((xmin, ymin), f"{labels[label_index]}: {round(float(score),2)}", fill="white")
    return img


with gr.Blocks(title="Zero-shot object detection", theme="freddyaboulton/dracula_revamped") as demo:
    gr.Markdown(""
            "## Zero-shot object detection"
            "")
    with gr.Row():
        with gr.Column():
            in_img = gr.Image(label="Input Image", type="pil")
            in_labels = gr.Textbox(label="Input labels, comma apart")
            inference_btn = gr.Button("Inference", variant="primary")
        with gr.Column():
            out_img = gr.Image(label="Result", interactive=False)
        
    inference_btn.click(inference, inputs=[in_img, in_labels], outputs=[out_img])

if __name__ == "__main__":
    demo.queue().launch()