how to read the image.list by pipeline about dynamic-training-bench

[galeone]

As you can see from the error

NotFoundError (see above for traceback): /Users/liuguiyang/Documents/CodeProj/PyProj/dtb/scripts/inputs/data/airplane/positive/32size/0335.png:1

The file /Users/liuguiyang/Documents/CodeProj/PyProj/dtb/scripts/inputs/data/airplane/positive/32size/0335.png:1 does not exist.

Check the content of image.list and look if the files listed there effectively exists.

(the :1 in the file name sounds strange to me, check this)

[liuguiyangnwpu]

hi,
after i check the imag.list, I can read the *.png images.

import os
import tensorflow as tf

filename = "/Users/liuguiyang/Documents/CodeProj/PyProj/dtb/scripts/inputs/data/airplane/positive/32size/0335.png"

def test02():
#    filename = "/Users/liuguiyang/Desktop/cifar/test/0_cat.png"
    with tf.Graph().as_default():
        image_contents = tf.read_file(filename)
        image = tf.image.decode_png(image_contents, channels=3)
        init_op = tf.tables_initializer()
        with tf.Session() as sess:
            sess.run(init_op)
            tmp = sess.run(image)
            print(tmp)
if __name__ == "__main__":
    test02()

the answer is below:

➜  dtb git:(CAE) ✗ python3 run.py > a.txt
➜  dtb git:(CAE) ✗ more a.txt 
[[[173 173 173]
  [173 173 173]
  [173 173 173]
  ..., 
  [176 176 176]
  [100 100 100]
  [117 117 117]]

 [[175 175 175]
  [174 174 174]
  [173 173 173]
  ..., 
  [177 177 177]
  [149 149 149]
  [136 136 136]]

 [[175 175 175]
  [173 173 173]
  [171 171 171]
  ..., 
  [173 173 173]
  [173 173 173]
  [161 161 161]]

 ..., 
 [[173 173 173]
  [173 173 173]
  [172 172 172]
  ..., 
  [174 174 174]
  [173 173 173]
  [173 173 173]]

 [[173 173 173]
  [173 173 173]
  [173 173 173]
  ..., 
  [152 152 152]
  [171 171 171]
  [170 170 170]]

 [[172 172 172]
  [172 172 172]
  [172 172 172]
  ..., 
  [146 146 146]
  [170 170 170]
  [173 173 173]]]

it really can read the image data.
I am very confused about this question ???

[galeone]

Your input pipeline was correct (the commented part, not the modified version you posted here with comments added and debug operations added), the only problem I see is in the image.list file content.

The image you read here does not have the :1 added at the end and the read operation worked well. Thus if every file listed in image.list exists everything should work fine

[liuguiyangnwpu]

hi,
i am very confused why my code is error?
But in another way, I get the same effect, below is my code.

import tensorflow as tf

filename_queue = tf.train.string_input_producer(
    tf.train.match_filenames_once("/Users/liuguiyang/Documents/CodeProj/PyProj/dtb/scripts/inputs/data/airplane/positive/32size/*.png")
)

image_reader = tf.WholeFileReader()
key, img_file = image_reader.read(filename_queue)
img = tf.image.decode_png(img_file)

with tf.Session() as sess:
    tf.global_variables_initializer().run()
    coord = tf.train.Coordinator()
    threads = tf.train.start_queue_runners(coord=coord)

    img_tensor = sess.run([img])
    print(img_tensor)

    coord.request_stop()
    coord.join(threads=threads)

thank you for your help ?