File size: 817 Bytes
99292ad 6dc9a2a 99292ad 6dc9a2a f0e891b 99292ad | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 | import gradio as gr
from ultralytics import YOLO
import spaces
@spaces.GPU
def predict_aerial_targets(input_img):
model = YOLO("best.pt")
# Load our custom-trained aerial model
# Run inference on the uploaded image
results = model(input_img, imgsz=640)
# Plot the bounding boxes directly onto the image
annotated_img = results[0].plot()
return annotated_img
# Define a simple drag-and-drop web interface
demo = gr.Interface(
fn=predict_aerial_targets,
inputs=gr.Image(type="pil", label="Upload Aerial/Satellite Image"),
outputs=gr.Image(type="numpy", label="Detected Targets"),
title="Nadir-Perspective Object Detection",
description="Drop a satellite or drone image here to detect planes, ships, and infrastructure."
)
if __name__ == "__main__":
demo.launch() |