Black Lives Matter - Action and Equality. ... Adafruit joins the Stop Hate for Profit campaign.
0

Tutorial: Running TensorFlow Lite Object Recognition on the
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

Tutorial: Running TensorFlow Lite Object Recognition on the

by geekguy on Thu Nov 07, 2019 5:37 pm

Hi,

I am working through the Running TensorFlow Lite Object Recognition on the Raspberry Pi 4 tutorial and have found a place where things do not work. In the Install TensorFlow 2.0 section, the URL that is used in wget (https://github.com/PINTO0309/Tensorflow-bin/raw/master/tensorflow-2.0.0rc0-cp37-cp37m-linux_armv7l.whl) gets me a 404 not found error. In fact, most of the URLs in the Tensorflow-bin directory of that user do not work.

Can an alternative source of these files be found? It looks like they have progressed to the released file state now.

8-Dale

geekguy
 
Posts: 271
Joined: Tue Sep 10, 2013 2:43 pm
Location: Beaverton, OR

Re: Tutorial: Running TensorFlow Lite Object Recognition on

by mikeysklar on Thu Nov 07, 2019 9:42 pm

Hello 8-Dale,

It looks like the TensorFlow archive had some files shuffled about this week. I think they dropped the rc0 (release candidate prefix). Here are some relevant links to try. Please let me know if this work:

High level overview of files:

https://github.com/PINTO0309/Tensorflow-bin

60MB RAW file link to the file you were looking for cp37m armv7l:
https://github.com/PINTO0309/Tensorflow ... l?raw=true

mikeysklar
 
Posts: 1199
Joined: Mon Aug 01, 2016 8:10 pm

Re: Tutorial: Running TensorFlow Lite Object Recognition on

by geekguy on Sun Nov 10, 2019 3:13 am

Hi,
mikeysklar wrote:Hello 8-Dale,

It looks like the TensorFlow archive had some files shuffled about this week. I think they dropped the rc0 (release candidate prefix). Here are some relevant links to try. Please let me know if this work:

High level overview of files:

https://github.com/PINTO0309/Tensorflow-bin

60MB RAW file link to the file you were looking for cp37m armv7l:
https://github.com/PINTO0309/Tensorflow ... l?raw=true

I have not tried your method, but here is what I had to do to get the right file to install:

Code: Select all | TOGGLE FULL SIZE
1) Clone the repository git@github.com:tensorflow/tensorflow.git
2) Go to the second page of the [url=https://learn.adafruit.com/running-tensorflow-lite-on-the-raspberry-pi-4/initial-setup]Running TensorFlow Lite Object Recognition on the Raspberry Pi 4[/url] tutorial, and do everything there. NOTE: You do NOT have to install the PiTFT Drivers if you do not have a PiTFT installed, because everything will run [b][i]fine[/i][/b] with an HDMI monitor.
3) On the third page of the tutorial, do everything down to, but [b]NOT[/b] including, the Install TensorFlow 2.0 section.
4)Go to https://github.com/PINTO0309/Tensorflow-bin/tensorflow-2.0.0-cp37-cp37m-linux_armv7l.whl in your browser
5) Click the Download button
6) cd to your copy of the tensorflow repository
7) pip3 install <path to the tensorflow-2.0.0-cp37-cp37m-linux_armv7l.whl file>/tensorflow-2.0.0-cp37-cp37m-linux_armv7l.whl
8) pip3 install -e .
9) Follow the rest of the tutorial as it is

8-Dale

geekguy
 
Posts: 271
Joined: Tue Sep 10, 2013 2:43 pm
Location: Beaverton, OR

Re: Tutorial: Running TensorFlow Lite Object Recognition on

by mountaintenor on Mon Dec 09, 2019 12:35 pm

I have also run into this problem, reported over a month ago. Could this tutorial be updated so that it works?

mountaintenor
 
Posts: 2
Joined: Sat Aug 04, 2012 10:42 am

Re: Tutorial: Running TensorFlow Lite Object Recognition on

by mikeysklar on Tue Dec 10, 2019 6:54 pm

Did you find this through an Adafruit guide? The tutorial itself on GitHub appears to be owned by Katsuya Hyodo a Japanese developer for Intel.

https://github.com/PINTO0309

mikeysklar
 
Posts: 1199
Joined: Mon Aug 01, 2016 8:10 pm

Re: Tutorial: Running TensorFlow Lite Object Recognition on

by mountaintenor on Wed Dec 11, 2019 10:45 am

I used the Adaftuit guide:

https://learn.adafruit.com/running-tens ... berry-pi-4

and got the 404 error.

mountaintenor
 
Posts: 2
Joined: Sat Aug 04, 2012 10:42 am

Re: Tutorial: Running TensorFlow Lite Object Recognition on

by mikeysklar on Fri Dec 13, 2019 6:58 pm

Gotcha. I updated the guide to reference the valid path for TensorFlow 2.0.

mikeysklar
 
Posts: 1199
Joined: Mon Aug 01, 2016 8:10 pm

Re: Tutorial: Running TensorFlow Lite Object Recognition on

by PINTO0309 on Sat Dec 14, 2019 7:54 pm

Hi, I am the author of Tensorflow v2 for RaspberryPi installer. Tensorflow upgrades are fast, so my repository top page is refreshed regularly. This is because Github has a 1GB capacity limit. It is safer to include this URL in the procedure as frequent link downs are problematic. All the latest installers and past installers can be downloaded from my GoogleDrive. The shell stored at this URL will not be removed in the future, so you don't have to worry about linking down. The reason for saving the installer on the top page of Github is that some engineers may not have access to GoogleDrive.
https://github.com/PINTO0309/Tensorflow-bin/tree/master/previous_versions

PINTO0309
 
Posts: 1
Joined: Sat Dec 14, 2019 7:44 pm

Re: Tutorial: Running TensorFlow Lite Object Recognition on

by mikeysklar on Sun Dec 15, 2019 6:52 pm

Katsuya-san,

Thank you for the response.

If you see any links in our TensorFlow guides you would like updated please let us know. Since we are referencing the WHL file I've not been able to use the previous_verions directory of shell scripts.

mikeysklar
 
Posts: 1199
Joined: Mon Aug 01, 2016 8:10 pm

Re: Tutorial: Running TensorFlow Lite Object Recognition on

by tbarton782 on Mon Mar 30, 2020 1:26 pm

I got the tutorial (https://learn.adafruit.com/teachable-ma ... a?view=all) to work, until literally the very last line of code ("python3 tests/pitft_teachablemachine.py ../converted_savedmodel.zip"). It errors out saying "python3: can't open file 'tests/pitft_teachablemachine.py': [Errno 2] No such file or directory".

This is a really cool project, and it would be fun to see it work. Did I fail to create a "tests/pitft_teachablemachine.py" file somewhere, or do I need to edit some text somewhere to use my own model?

Thoughts?

Thank you, Tyler
Attachments
Capture.JPG
Capture.JPG (31.61 KiB) Viewed 27 times

tbarton782
 
Posts: 1
Joined: Mon Mar 30, 2020 1:05 pm

Please be positive and constructive with your questions and comments.