Not able to run monodepth_simple: Unsuccessful TensorSliceReader constructor about monodepth HOT 7 CLOSED

SUCCESS!

A very simple solution for me - I don't understand how it makes a difference and I couldn't find this problem/solution listed anywhere else: I just used absolute paths for all files, so the --image_path and also the --checkpoint_path flags.

from monodepth.

Hello,
I am trying to make the same network work for me but I get this error,
" Traceback (most recent call last):
File "monodepth_simple.py", line 110, in
tf.app.run()
File "/opt/conda/lib/python3.5/site-packages/tensorflow/python/platform/app.py", line 125, in run
_sys.exit(main(argv))
File "monodepth_simple.py", line 107, in main
test_simple(params)
File "monodepth_simple.py", line 76, in test_simple
train_saver.restore(sess, restore_path)
File "/opt/conda/lib/python3.5/site-packages/tensorflow/python/training/saver.py", line 1717, in restore
+ compat.as_text(save_path))
ValueError: The passed save_path is not a valid checkpoint: "

DO you have any idea about this?

from monodepth.

UPDATE: I found on another thread that you did the work using Python 2.7 (along with the corresponding package version, handled using conda environments).
Upgrading to that got me over the above mentioned error. Now I am getting another error.

This appears many times:
tensorflow/core/framework/op_kernel.cc:993] Not found: Unsuccessful TensorSliceReader constructor: Failed to find any matching files for W

Then finally this:

NotFoundError (see above for traceback): Unsuccessful TensorSliceReader constructor: Failed to find any matching files for
[[Node: save/RestoreV2_8 = RestoreV2[dtypes=[DT_FLOAT], _device="/job:localhost/replica:0/task:0/cpu:0"](_recv_save/Const_0, save/RestoreV2_8/tensor_names, save/RestoreV2_8/shape_and_slices)]]

Does anybody know where I should start looking for a fix? Have a misspecified a parameter, or should it really trying to restore things?

from monodepth.

def scale_pyramid(self, img, num_scales):
scaled_imgs = [img]
s = tf.shape(img)
h = s[1]
w = s[2]
for i in range(num_scales - 1):
ratio = 2 ** (i + 1)
nh = h // ratio ####
nw = w // ratio #### @nicholas-mitchell
scaled_imgs.append(tf.image.resize_area(img, [nh, nw]))
return scaled_imgs

from monodepth.

@dleam - could you maybe say a few words to your comment? I don't understand how that helps...

from monodepth.

Further attempts to get the simple model working:

I have switched to tensorflow-gpu version 1.0.1 (hadn't thought it would be necessary for the simple model test). This still produces the same error, albeit after recognising my graphics card as the device. This came with cudatoolkit version 7.xx - so I then updated that to cudatoolkit version 8 as used by the authors. This automatically upgraded tf to version 1.2.11 and uses CUDNN 5.1. These also still produce the same error as above.

from monodepth.

Hello,
I am trying to make the same network work for me but I get this error,
" Traceback (most recent call last):
File "monodepth_simple.py", line 110, in
tf.app.run()
File "/opt/conda/lib/python3.5/site-packages/tensorflow/python/platform/app.py", line 125, in run
_sys.exit(main(argv))
File "monodepth_simple.py", line 107, in main
test_simple(params)
File "monodepth_simple.py", line 76, in test_simple
train_saver.restore(sess, restore_path)
File "/opt/conda/lib/python3.5/site-packages/tensorflow/python/training/saver.py", line 1717, in restore

• compat.as_text(save_path))
  ValueError: The passed save_path is not a valid checkpoint: "

DO you have any idea about this?

I solved this problem by using the absolute path.

from monodepth.