app.py

import gradio as gr
from transformers import pipeline

pipeline = pipeline(task="image-classification", model="bortle/astrophotography-object-classifier-alpha5")

def predict(image):
  predictions = pipeline(image)
  return {p["label"]: p["score"] for p in predictions}


def process_image(image):
    width = 1080
    ratio = width / image.width
    height = int(image.height * ratio)
    resized_image = image.resize((width, height))
    return resized_image

gr.Interface(
    predict,
    fn=process_image,
    inputs=gr.Image(type="pil", label="Upload Astrophotography image"),
    outputs=gr.Label(num_top_classes=5),
    title="Astrophotography Object Classifier",
    allow_flagging="manual",
    examples=["examples/Andromeda.jpg", "examples/Heart.jpg", "examples/Pleiades.jpg", "examples/Rosette.jpg", "examples/Moon.jpg", "examples/GreatHercules.jpg", "examples/Leo-Triplet.jpg", "examples/Crab.jpg", "examples/North-America.jpg", "examples/Horsehead-Flame.jpg", "examples/Pinwheel.jpg", "examples/Saturn.jpg"],
    cache_examples=True
).launch()