main_classify_fish.py

import cv2
import numpy as np
from tensorflow.keras.preprocessing.image import load_img
from tensorflow.keras.preprocessing.image import img_to_array
from tensorflow.keras.applications.mobilenet_v2 import preprocess_input
from tensorflow.keras.applications import MobileNetV2
from tensorflow.keras.models import Model
from tensorflow.keras.layers import Input, Dense
from tensorflow.keras.layers import GlobalAveragePooling2D 

fish_classes = ['Polyprion americanus', 'Scomber japonicus',
       'Trigloporus lastoviza', 'Dasyatis centroura',
       'Tetrapturus belone', 'Anthias anthias', 'Coris julis',
       'Solea solea', 'Boops boops', 'Chlorophthalmus agassizi',
       'Squalus acanthias', 'Trachinus draco', 'Pseudocaranx dentex',
       'Belone belone', 'Gobius niger', 'Mugil cephalus', 'Phycis phycis',
       'Atherinomorus lacunosus', 'Rhinobatos cemiculus',
       'Epinephelus caninus']


def model_initalize():
    # include_top = False means that we doesnt include fully connected top layer we will add them accordingly
    mobilenetV2 = MobileNetV2(include_top = False, input_shape = (224,224,3), weights = 'imagenet')

    # training of all the convolution is set to false
    for layer in mobilenetV2.layers:
        layer.trainable = False

    x = GlobalAveragePooling2D()(mobilenetV2.output)    
    predictions = Dense(20, activation='softmax')(x)
    model_mobilenetV2 = Model(inputs = mobilenetV2.input, outputs = predictions)
    model_mobilenetV2.load_weights('model_mobilenetV2.h5')
    return model_mobilenetV2

model = model_initalize()

def getPrediction(image):
    image = img_to_array(image)
    image = cv2.resize(image, (224,224))
    image = image.reshape((1, image.shape[0], image.shape[1], image.shape[2]))
    image = preprocess_input(image)
    probs = model.predict(image)
    yhat = np.argmax(probs)
    return str(fish_classes[yhat]), str(probs[0][yhat])