2023 Undergraduate Conference Model Architecture - UMIATR - 2023 - York University - License to View, NOT use commercially

model.py

import os.path
import shutil
from glob import glob
from typing import Optional, List

from keras import Sequential, Model
from keras.callbacks import History, Callback, ModelCheckpoint
from keras.layers import Conv2D, Dropout, MaxPooling2D, Flatten, Dense, BatchNormalization, Activation, Rescaling
from matplotlib import pyplot
import pandas as pd


class RATModelClient:
    """
    Class to manage the RAT Model

    """

    @classmethod
    def create(cls) -> Sequential:
        """
        Create the un-built version of the RAT CNN Model

        :return: Sequential model

        """

        model: Sequential = Sequential([

            Rescaling((1. / 255), input_shape=(256, 256, 1)),

            Conv2D(32, (3, 3), padding='same'),
            BatchNormalization(),
            Activation('relu'),
            MaxPooling2D(pool_size=(2, 2), strides=(2, 2)),

            Conv2D(64, (3, 3), padding='same'),
            BatchNormalization(),
            Activation('relu'),
            MaxPooling2D(pool_size=(2, 2), strides=(2, 2)),

            Conv2D(128, (3, 3), padding='same'),
            BatchNormalization(),
            Activation('relu'),
            MaxPooling2D(pool_size=(2, 2), strides=(2, 2)),

            Flatten(),
            Dense(512),
            BatchNormalization(),
            Activation('relu'),
            Dropout(0.5),

            Dense(2, activation='sigmoid')

        ])

        model.build()
        return model