Code for my medium story Serve a Deep Learning Image Classification Model Written in TensorFlow 2 as a REST API and Dockerize it! (with GPU support)

read_image.py

def read_image(image_path, image_size):
    """ read an image using opencv """
    
    image = cv2.imread(image_path) #BGR
    # convert to RGB
    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    # resize using cubic interpolation, a way to deal with missing data (upsize or downsize)
    image = cv2.resize(image, image_size, cv2.INTER_CUBIC) 
    
    return image

def load_dataset(lables_file_path, dataset_path, image_size):
    """ load data """
    
    with open(lables_file_path,  "r") as f:
        labels = f.read().splitlines() # ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']
        
    images = []
    targets = []

    for image_file_name in tqdm(os.listdir(dataset_path), desc="loading images", unit="image loading"):
        
        try:
            image_path = os.path.join(dataset_path, image_file_name)
            images.append(image_loader(image_path, image_size))
            
            for idx in range(len(labels)):
                if labels[idx] in image_file_name: #image file name contains its label
                    targets.append(idx)
        except Exception as e:
            pass
       
    return np.array(images), np.array(targets)