通过 U-Net 进行图像分割时出现形状错误：ValueError：无法将输入数组从形状 （128,128） 广播到形状 （128,128,3）

问：

我正在尝试使用 U-Net 开发皮肤分割。我在下面的代码块中遇到了一个错误，无法解决它。我到底在哪里犯了错误？ 守则，

for n, id_ in tqdm(enumerate(os.listdir(TRAIN_PATH)), total=len(os.listdir(TRAIN_PATH))):
    path = TRAIN_PATH
    img = imread(path + id_)[:, :, :IMG_CHANNELS]
    img = resize(img, (IMG_HEIGHT, IMG_WIDTH), mode='constant', preserve_range=True)
    X_train[n] = img
    mask = np.zeros((IMG_HEIGHT, IMG_WIDTH, 1), dtype=np.bool_)
    for root, dirs, files in os.walk(path + '/masks/'):
        for mask_file in files:
            mask_ = imread(os.path.join(root, mask_file))
            mask_ = np.expand_dims(resize(mask_, (IMG_HEIGHT, IMG_WIDTH), mode='constant',
    preserve_range=True), axis=-1)
            mask = np.maximum(mask, mask_)

    Y_train[n] = mask

错误，

Traceback (most recent call last):
  File "..\main.py", line 58, in <module>
    X_test[n] = img
ValueError: could not broadcast input array from shape (128,128) into shape (128,128,3)