app.py

# import gradio as gr
# import tensorflow as tf
# from PIL import Image
# import numpy as np

# # Teachable Machineで作成したモデルのパス
# model_path = "keras_model.h5"

# # モデルの読み込み
# model = tf.keras.models.load_model(model_path)

# # クラスのラベル（Teachable Machineで指定したクラス名と同じ順序で設定）
# class_labels = [
# "Cotton",
# "Linen",
# "Silk",
# "Wool",
# "Polyester",
# "Nylon",
# "Rayon",
# "Fleece",
# "Leather",
# "Synth Leather"
# ]

# # GradioのUIの設定
# def classify_image(img):
#     # 画像の前処理
#     img = Image.fromarray((img * 255).astype(np.uint8))
#     img = img.resize((224, 224))
#     img_array = tf.keras.preprocessing.image.img_to_array(img)
#     img_array = tf.expand_dims(img_array, 0)  # バッチの次元を追加

#     # モデルの予測
#     predictions = model.predict(img_array)
#     predicted_class = class_labels[np.argmax(predictions)]

#     return predicted_class

# # Gradio UIの作成
# iface = gr.Interface(
#     fn=classify_image,
#     inputs=gr.Image(),
#     outputs=gr.Textbox(),
#     # live=True,
# )

# # UIの起動
# iface.launch(share=True)
from fastai.vision.all import *
import gradio as gr

learn = load_learner('export.pkl')

categories = ('Blouse', 'Dress', 'Pants', 'Shirt', 'Shorts')
title = "Clothing Identifier"

def classify_image(img):
    pred, idx, probs = learn.predict(img)
    highest_prob_index = probs.argmax()  # Find the index of the highest probability
    return categories[highest_prob_index]  # Return the class with the highest probability



iface = gr.Interface(
    fn=classify_image,
    inputs=gr.Image(),
    outputs=gr.Textbox(),
    title = title,
    examples = ['dress.jpg', 'shirt.jpg', 'pants.jpg', 'shorts.jpg'],
    
    # live=True,
)
iface.launch(share=True)
# image = gr.Image(shape=(512, 512))
# label = gr.Label()
# examples = ['dress.jpg', 'shirt.jpg', 'pants.jpg', 'shorts.jpg']

# intf = gr.Interface(fn=classify_image, title=title, inputs=image, outputs=label, examples=examples)
# intf.launch(inline=False)