import spaces
import gradio as gr
import cv2
import tempfile
from ultralytics import YOLOv10

model = YOLOv10.from_pretrained(f'jameslahm/yolov10s')

@spaces.GPU
def yolov10_inference(image, conf_threshold):
    width, _ = image.size
    results = model.predict(source=image, imgsz=width, conf=conf_threshold)
    annotated_image = results[0].plot()
    return annotated_image[:, :, ::-1]


def app():
    with gr.Blocks():
        with gr.Row():
            with gr.Column():
                image = gr.Image(type="pil", label="Image", visible=True)
                conf_threshold = gr.Slider(
                    label="Confidence Threshold",
                    minimum=0.0,
                    maximum=1.0,
                    step=0.05,
                    value=0.25,
                )

            with gr.Column():
                output_image = gr.Image(type="numpy", label="Annotated Image", visible=True)

        image.stream(
            fn=yolov10_inference,
            inputs=[image, conf_threshold],
            outputs=[image],
            stream_every=0.1,
            time_limit=30
        )


gradio_app = gr.Blocks()
with gradio_app:
    gr.HTML(
        """
    <h1 style='text-align: center'>
    YOLOv10: Real-Time End-to-End Object Detection
    </h1>
    """)
    gr.HTML(
        """
        <h3 style='text-align: center'>
        <a href='https://arxiv.org/abs/2405.14458' target='_blank'>arXiv</a> | <a href='https://github.com/THU-MIG/yolov10' target='_blank'>github</a>
        </h3>
        """)
    with gr.Row():
        with gr.Column():
            app()
if __name__ == '__main__':
    gradio_app.launch()