ModuleNotFoundError: No module named 'tensorflow.models'

[rs1990]

Hi,

I’m new to Tensorflow and am trying to test this package. but i keep getting this error.

File "image_recognition.py", line 9, in
from tensorflow.models.image.imagenet import classify_image
ModuleNotFoundError: No module named 'tensorflow.models'

Id appreciate any help.
thanks

RS

[OTL]

Hi, thank you for feedback!

What is the version of ternsorflow?
This is created for v0.9, i guess that it is too old.

If you can install v0.9, how about trying it first?

If you can fix the library path for newest tensorflow, PR is welcome!

[rs1990]

im using version 1.2.1.

I tried fixing that but i get this issue now :

raghavendra_sriram@mav:~/indigo_ws/src/rostensorflow$ python image_recognition.py image:=/cv_camera/image_raw
Traceback (most recent call last):
  File "image_recognition.py", line 54, in <module>
    tensor = RosTensorFlow()
  File "image_recognition.py", line 19, in __init__
    classify_image.maybe_download_and_extract()
  File "/home/raghavendra_sriram/indigo_ws/src/models/tutorials/image/imagenet/classify_image.py", line 172, in maybe_download_and_extract
    dest_directory = FLAGS.model_dir
AttributeError: 'NoneType' object has no attribute 'model_dir'

[rs1990]

Are there any other necessary requirements to run this package ?

[OTL]

Hi, newest classify_image.py seems that it needs some modification. So I copied it in my repo, and
modified to use from my program.

And I push the new version, which supports kinetic and latest tensorflow for you!!!
Please update, and try it again!

[rs1990]

Hello,
Thanks a lot. Sorry i couldnt get back to you earlier. I managed to get the previous version working with tensorflow 0.9. Il try this with tensorflow 1.0 and get back.

Cheers.

[OTL]

don't mind, and you don't need to be harry. 2017/08/09 8:07 "Raghavendra Sriram" <[email protected]>: Hello, Thanks a lot. Sorry i couldnt get back to you earlier. I managed to get the previous version working with tensorflow 0.9. Il try this with tensorflow 1.0 and get back. Cheers. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAMpJkmLDETvvuPpk94qPok6d1qYBXxfks5sWOo2gaJpZM4OmM-t> .