app.py

import numpy as np
import tensorflow as tf
import gradio as gr
from huggingface_hub import hf_hub_download
from PIL import Image
import json

model_path = hf_hub_download(
    repo_id="tancnle/smart-recycling",
    filename="model_baseline.h5",
    use_auth_token="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",
)

dim = (299, 299)


def read_image(image):
    image = tf.convert_to_tensor(image)
    image.set_shape([None, None, 3])
    image = tf.image.resize(images=image, size=dim)
    image = image / 125.0 - 1
    return image


def infer(model, image_tensor):
    predictions = model.predict(tf.expand_dims(image_tensor, axis=0))
    labels = ["cardboard", "glass", "metal", "paper", "plastic", "trash"]
    predictions = list(map(float, predictions[0]))
    output = dict(zip(labels, predictions))

    return output


def top_3_accuracy(ytrue, ypred):
    return tf.keras.metrics.sparse_top_k_categorical_accuracy(ytrue, ypred, k=3)


def classify(input_image):
    model = tf.keras.models.load_model(
        model_path,
        custom_objects={"top_3_accuracy": top_3_accuracy},
    )

    image_tensor = read_image(input_image)
    predictions = infer(model, image_tensor)
    return predictions


title = "Classify Trash"
description = "Upload an image or select from examples to classify trash."
article = "<div style='text-align: center;'>Space by Tan Le</div>"

examples = [
    "images/cardboard.jpeg",
    "images/cigarette_butt.jpeg",
    "images/masks.jpeg",
    "images/metal_objects.jpeg",
    "images/paper.jpeg",
    "images/plastic.jpeg",
    "images/spray_cans.jpeg",
    "images/syringe.jpeg",
]

demo = gr.Interface(
    classify,
    inputs=gr.inputs.Image(),
    outputs="label",
    examples=examples,
    title=title,
    description=description,
    article=article,
)
demo.launch()