-
Notifications
You must be signed in to change notification settings - Fork 70
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Error encountered when trying to use Google Coral accelerator #37
Comments
Thanks for the bug report @GBW48, sorry this didn't work for you! Looks like your TensorFlow installation didn't finish because you don't have the
$ echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list
$ curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -
$ sudo apt-get update
$ sudo apt-get install libedgetpu1-std
|
Hi Leigh, Thanks so much for the quick response. With your suggested changes, I am still getting run-time errors and I'm unable to display objects using the Coral accelerator. I tried your suggestions on a fresh build. The wheel and TensorFlow installation appeared to work. Here are the results of the install.
The install of Coral's libedgetpu appears to have errors. Here are the results of the install.
The install or Coral's tflite_runtime appeared to work. Here are the results of the install.
I confirmed that object detection using the "rpi-deep-pantilt detect --rotation 180" command worked.
Here is the console log when I tried to use the Coral accelerator.
Thanks in advance. I really appreciate your help with this issue... |
Hi Leigh, I just wanted to let you know that I tried your fix for the Coral runtime error. Unfortunately, the suggested changes did not work. I posted the results on GitHub.
When you have an opportunity, I’d appreciate it if you could look into the Coral runtime issue. Thanks in advance! ...Gary White
…------ Original Message ------
From: Leigh Johnson
To: leigh-johnson/rpi-deep-pantilt
Cc: GBW48, Mention
Sent: June 24, 2020 at 10:53 PM
Subject: Re: [leigh-johnson/rpi-deep-pantilt] Error encountered when trying to use Google Coral accelerator (#37)
Thanks for the bug report @GBW48 (https://github.com/GBW48), sorry this didn't work for you!
Looks like your TensorFlow installation didn't finish because you don't have the wheel Python package installed. I think my Coral installation instructions are out of date too. Try the following to re-install, let me know how it goes. 🤞
Let me know how it goes a (with your virtual environment active) to reinstall everything
Install wheel and reinstall TensorFlow
(.venv)$ pip install wheel && pip install --force-reinstall https://github.com/leigh-johnson/Tensorflow-bin/releases/download/v2.2.0/tensorflow-2.2.0-cp37-cp37m-linux_armv7l.whl
Install/reinstall Coral's libedgetpu.
$ echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list $ curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add - $ sudo apt-get update $ sudo apt-get install libedgetpu1-std
Re-install Coral's tflite_runtime
(.venv)$ pip install https://dl.google.com/coral/python/tflite_runtime-2.1.0.post1-cp37-cp37m-linux_armv7l.whl
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub (#37 (comment)), or unsubscribe (https://github.com/notifications/unsubscribe-auth/AJ6UNZBY42SPR5E2MIFTGMTRYK33PANCNFSM4OHKDX7Q).
|
Hi Leigh, I reran your instructions for the install/reinstall of Coral's libedgetpu. This time there no errors. After a reboot, I was able to successfully use the Coral accelerator to display and track both objects and faces. Every thing worked! Thanks so much for creating such an excellent, well conceived and documented project! What a wonderful way to learn about the capabilities of TensorFlow! P.S. (.venv) pi@raspberrypi:~ $ echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list
(.venv) pi@raspberrypi:~ $ curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -
(.venv) pi@raspberrypi:~ $ sudo apt-get update
(.venv) pi@raspberrypi:~ $ sudo apt-get install libedgetpu1-std
I then re-installed Coral's tflite_runtime which also worked. Here are the results:
|
Sorry, I closed this issue by mistake. |
@GBW48 Woooo, glad that worked for you Gary! Thanks for including logs of your second go. At a glance, I'm not sure what's different. 🤔 I'll prioritize getting installation a bit better scripted/automated. Right now, there are a lot of manual steps that make it tedious to reproduce a working environment. I've been getting by with some Ansible playbooks for my own release testing, but they're too crufty and personal to release for general use. |
Description
Object detection using Google Coral accelerator (following README.md instructions for rpi-deep-pantilt version 1.2) fails and does not display image. Please note standard object detection without using the accelerator works ("rpi-deep-pantilt detect --rotation 180"). I repeated the deep-pan-tilt build and object detection twice -- with the same result.
When I issue "rpi-deep-pantilt detect --edge-tpu --rotation 180", I get two errors with no image displayed on the RPi.
Fist error:
Second error:
What I Did
I freshly installed Raspbian 10 (buster) on a high speed 32 gig SD card. I then followed the README.md instructions. I only had one issue after issuing the "pip install https://github.com/leigh-johnson/Tensorflow-bin/releases/download/v2.2.0/tensorflow-2.2.0-cp37-cp37m-linux_armv7l.whl" command.
Here is the error:
Otherwise, I encountered no problems with the install. Please note, I was careful to make all recommended changes config.txt and Raspberry Pi interfaces.
The text was updated successfully, but these errors were encountered: