python - 当我尝试在 Tensorflow 中调整图像大小时如何修复 "TypeError: x and y must have the same dtype, got tf.uint8 != tf.float32"

Question

我尝试设置一个图像管道，为 Tensorflow 构建图像数据集来裁剪图像，但我无法裁剪图片。 我关注了这个tutorial但我想将文件裁剪为正方形，并且在不保留纵横比的情况下不调整其大小。
这是我的代码:

#
import tensorflow as tf
#

img_raw = tf.io.read_file('./images/4c34476047bcbbfd10b1fd3342605659.jpeg/')

img_tensor = tf.image.decode_jpeg(img_raw, channels=3)

img_final = tf.image.crop_to_bounding_box(
    img_tensor,
    0,
    0,
    200,
    200
)

img_final /= 255.0  # normalize to [0,1] range

当我像教程中那样使用简单的图像调整大小时，它会起作用:

#
import tensorflow as tf
#

img_raw = tf.io.read_file('./images/4c34476047bcbbfd10b1fd3342605659.jpeg/')

img_tensor = tf.image.decode_jpeg(img_raw, channels=3)

img_final = tf.image.resize(img_tensor, [192, 192])

img_final /= 255.0  #

这是日志:

    img_final /= 255.0  # normalize to [0,1] range
  File ".../.local/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py", line 876, in binary_op_wrapper
    return func(x, y, name=name)
  File ".../.local/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py", line 964, in _truediv_python3
    (x_dtype, y_dtype))
TypeError: x and y must have the same dtype, got tf.uint8 != tf.float32

Answer 1

Tensorflow 不进行自动类型转换。看来解码后的图像是一个 dtype 为 tf.uint8 的张量，您必须将其转换为 tf.float32 才能使除法按您的预期工作。

替换

img_final/= 255.0

与

img_final = tf.cast(img_final, tf.float32)/255.0