Python深度学习调试，常用操作：Real-time Update...

1.查看数据类型：

type(object)

2.查看数据维度:

object.shape

3. 图像读取，解码，改变大小操作

报错：Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).

image_string  = tf.read_file(name)

image = tf.image.decode_jpeg(image_string) # 返回的是uint8

image = tf.image.convert_image_dtype(image,tf.float32)#转换数据类型
image = tf.image.resize_images(image, (224,224)) #返回的是float

或者float类型的数据：image = image / 255

float要在0-1之间，int要在0-255之间

4. tensor和数组之间的转换：

import tensorflow as tf
img1 = tf.constant(value=[[[[1],[2],[3],[4]],[[1],[2],[3],[4]],[[1],[2],[3],[4]],[[1],[2],[3],[4]]]],dtype=tf.float32)
img2 = tf.constant(value=[[[[1],[1],[1],[1]],[[1],[1],[1],[1]],[[1],[1],[1],[1]],[[1],[1],[1],[1]]]],dtype=tf.float32)
img = tf.concat(values=[img1,img2],axis=3)      #<class 'tensorflow.python.framework.ops.Tensor'>
sess=tf.Session()

#sess.run(tf.initialize_all_variables())
sess.run(tf.global_variables_initializer())

#转化为numpy数组
img_numpy=img.eval(session=sess) #<class 'numpy.ndarray'>

#转化为tensor
img_tensor= tf.convert_to_tensor(img_numpy) #<class 'tensorflow.python.framework.ops.Tensor'>

5. Failed to get convolution algorithm. This is probably because cuDNN failed to initialize

tensorflow:

config = tf.ConfigProto()

config.gpu_options.allow_growth = True

with tf.Session(config=config) as sess:

keras:

config = tf.ConfigProto()

config.gpu_options.allow_growth = True

keras.backend.tensorflow_backend.set_session(tf.Session(config=config))

6.正常的Anaconda深度学习环境搭建报错：

可能是Anaconda的安装目录被隐藏，取消隐藏即可。