??????????

1. ?????????

???????????????????????????????????????????

????????????????????????????????????????2????????????????????????????

2. ??????

???????????????????????????

import sys
sys.path.append(r'E:\Pycharm\project\yeah&ok\load_data')  
from load_data import load_data_func, test_image, augment

??????load_data_func?test_image???

3. ?????????

?????????????????????????????

data_dir = 'E:\Pycharm\project\yeah&ok\dataset'
Batch_size = 32  # ?????
train_dataset, test_dataset = load_data_func(data_dir, batchsize=Batch_size)
test_image(train_dataset)  # ??9???

????????????????????????

4. ?????

????????

import tensorflow as tf
import matplotlib.pyplot as plt
import numpy as np
import pathlib
import random
import tensorflow_datasets as tfds

5. ?????

??????????????????????????????

def load_data_func(data_dir, batch_size):
    data_root = pathlib.Path(data_dir)
    all_image_path = list(data_root.glob('*/*'))
    all_image_path = [str(path) for path in all_image_path]
    random.shuffle(all_image_path)
    
    label_names = sorted(item.name for item in data_root.glob('*/') if item.is_dir())
    label_to_index = {name: idx for idx, name in enumerate(label_names)}
    print(label_to_index)
    
    all_image_label = [label_to_index[pathlib.Path(p).parent.name] for p in all_image_path]
    print(len(all_image_label))
    
    image_path = all_image_path[5]
    image_show = (1 + load_preprocess_image(image_path)) / 2.
    plt.imshow(image_show)
    plt.show()
    
    path_ds = tf.data.Dataset.from_tensor_slices(all_image_path)
    image_dataset = path_ds.map(load_preprocess_image)
    label_dataset = tf.data.Dataset.from_tensor_slices(all_image_label)
    dataset = tf.data.Dataset.zip((image_dataset, label_dataset))
    
    image_count = len(all_image_path)
    test_count = int(image_count * 0.2)
    train_count = image_count - test_count
    print(test_count, train_count)
    
    train_dataset = dataset.skip(test_count).shuffle(buffer_size=150).repeat(3).batch(batch_size)
    test_dataset = dataset.take(test_count).batch(batch_size)
    
    train_dataset = train_dataset.map(augment)
    return train_dataset, test_dataset

6. ???????

???????????????

def load_preprocess_image(img_path):
    img_raw = tf.io.read_file(img_path)
    img_tensor = tf.image.decode_jpeg(img_raw, channels=3)
    img_tensor = tf.image.resize(img_tensor, [160, 160])
    img_tensor = tf.cast(img_tensor, tf.float32)
    img = img_tensor / 127.5 - 1
    return img

7. ??????

???????????

def augment(image, label):
    image = tf.image.random_flip_left_right(image)
    image = tf.image.random_contrast(image, lower=0.0, upper=1.0)
    image = tf.image.random_flip_up_down(image)
    image = tf.image.random_brightness(image, max_delta=0.5)
    image = tf.image.random_hue(image, max_delta=0.3)
    image = tf.image.random_saturation(image, lower=0.3, upper=0.5)
    return image, label

8. ??????

???????????

def test_image(train_dataset):
    plt.figure(figsize=(12, 12))
    for batch in tfds.as_numpy(train_dataset):
        for i in range(9):
            image, label = (1 + batch[0][i]) / 2., batch[1][i]
            plt.subplot(3, 3, i + 1)
            plt.imshow(image)
            plt.grid(False)
        break
    plt.show()