app.py

import gradio as gr
import cv2
import requests
import json
from PIL import Image
import numpy as np
import os
import gradio as gr
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing import image
from tensorflow.keras.applications.mobilenet_v2 import preprocess_input
import numpy as np

model = load_model('artikel.h5')


def preprocess_image(img):
    img = np.array(img)  # Ensure img is a numpy array
    img = cv2.resize(img, (224, 224))  # Resize using cv2 which is already imported
    img = np.expand_dims(img, axis=0)  # Expand dims to add the batch size
    return preprocess_input(img)  # Use the MobileNet-specific preprocessing

def predict_image(img):
    processed_image = preprocess_image(img)
    prediction = model.predict(processed_image)
    predicted_class_index = np.argmax(prediction, axis=1)[0]
    return predicted_class_index

def process_image(image):
    predicted_class_index=predict_image(image)
    with open('artikel.json', 'r') as file:
        data = json.load(file)
        
    predicted_class_index=str(data.get(str(predicted_class_index),"-1"))
    return(predicted_class_index)
    
interface = gr.Interface(
    fn=process_image,
    inputs=[gr.Image(type="numpy")],
    outputs=[gr.Textbox(label="Prediction")]
)

interface.launch(share=True)