我当前正在调试问题，并将其简化为这两个// select all DOM elements of type 'input' with css class 'radio-2' that are checked!
if ($('input.radio-2 :checked').size() > 0) {
//call function 1
} else {
//call function 2
}
循环，这些循环将输入张量的一部分切成小段并将其写入while。但是我得到的值从未在输入和切片中出现（例如TensorArray）。
这是一个最小的示例（在[[[900.]]]]和tf.while_loop上有两个x）：
y我不确定是什么问题； def demo(input):
# Shapes
input_shape = tf.shape(input)
i_height = input_shape[1]
i_width = input_shape[2]
i_c = input_shape[3]
ta = tf.TensorArray(dtype=input.dtype, size=i_height*i_width)
def body1(i, ta):
x = tf.constant(0)
condition2 = lambda x, i, result: tf.less(x, i_height)
def body2(x, y, ta):
sl = input[:, x:(x+1), y:(y+1)]
return x + 1, y, ta.write(index = x * i_height + y, value = sl)
x, i, ta = tf.while_loop(condition2, body2, loop_vars=[x, i, ta])
return i+1, ta
y = tf.constant(0)
condition1 = lambda y, result: tf.less(y, i_width)
y, ta = tf.while_loop(condition1, body1, [y, ta])
return ta.stack()
#return tf.reshape(ta.stack(), [-1, i_height, i_width, i_c])
HEIGHT = 10
WIDTH = 10
IN_CHANNELS = 1
with tf.Session() as sess:
imgs = tf.placeholder(tf.float32, [None, HEIGHT, WIDTH, IN_CHANNELS], name="imgs")
imgs_random = np.arange(10 * HEIGHT * WIDTH * IN_CHANNELS).reshape((10, HEIGHT, WIDTH, IN_CHANNELS)).astype('f')
feed_dict={imgs: imgs_random}
res = demo(imgs)
sess.run(tf.global_variables_initializer())
value = sess.run(res, feed_dict=feed_dict)#, options=options, run_metadata=run_metadata)
print("result", value.shape)
print("value", value)
print("rand", imgs_random[0])
的文档非常缺乏。