models.py

import tensorflow as tf
import numpy as np
import cv2


def outputs(y):
    return {
        "Achaemenid architecture": y[0],
        "American craftsman style": y[1],
        "American Foursquare architecture": y[2],
        "Ancient Egyptian architecture": y[3],
        "Art Deco architecture": y[4],
        "Art Nouveau architecture": y[5],
        "Baroque architecture": y[6],
        "Bauhaus architecture": y[7],
        "Beaux Arts architecture": y[8],
        "Byzantine architecture": y[9],
        "Chicago school_architecture": y[10],
        "Colonial architecture": y[11],
        "Deconstructivism": y[12],
        "Edwardian architecture": y[13],
        "Georgian architecture": y[14],
        "Gothic architecture": y[15],
        "Greek Revival architecture": y[16],
        "International style": y[17],
        "Novelty architecture": y[18],
        "Palladian architecture": y[19],
        "Postmodern architecture": y[20],
        "Queen Anne architecture": y[21],
        "Romanesque architecture": y[22],
        "Russian Revival_architecture": y[23],
        "Tudor Revival architecture": y[24],
    }


def efficientnetv2b0_25_arch_styles_Classifier(image):
    # file_path = f"./images/{file.filename}"
    # with open(file_path, "wb") as f:
    #     f.write(file.file.read())
    resized_image = cv2.resize(image, dsize=(
        224, 224), interpolation=cv2.INTER_CUBIC)
    img = tf.expand_dims(resized_image, 0)
    efficientnetv2b0 = tf.keras.models.load_model(
<<<<<<< HEAD
        "models\EfficientNetV2B0.h5")

    y = efficientnetv2b0.predict(img).reshape(-1)
    y = (np.round(y, 3)).tolist()
=======
        "EfficientNetV2B0.h5")

    y = efficientnetv2b0.predict(img).reshape(-1)
    y = (np.round(y, 3)*100).tolist()
>>>>>>> 6dfb3eb853ff6baf202a3117ebb5b61755454768

    return outputs(y)