devboard | interactive visual feedback tool for developing user | Frontend Framework library

I've been reading a lot about Jetson Nano and Google Coral Devboard and in most documentation and papers i've read, the inferencing and deployment are done using prebuilt convolutional neural networks such as AlexNet, Inception, MobileNet and other neural networks used for image classification. From what i understand these microcomputers require that the neural network is converted to a tensorflow model or any framework they accept to perform inferencing of the model.

What i would like to know is: for both Jetson Nano and Google Coral Devboard, can i have my own convolutional neural network that has nothing to do with those convolutional neural networks examplified in the documentation and deploy them to those boards?

I have been using the Google Coral Devboard in a project for a while. Today, I am surprised to see that using ssh does not work. Something went wrong somewhere. ssh mendel@192.168.0.111 gives

ssh: connect to host 192.168.0.111 port 22: Connection refused

mdt devices does find the device though. mdt shell does not work due to the inherent failure of ssh connection.

I have checked the authorized_keys file and my own private keys and they match.

I can only connect to the devboard using Serial communication. I wish to reflash the entire thing and reset the board. However, after using sudo reboot-bootloader, fastboot devices sees nothing.

I have tried 4 USB C cables. I have tried using fastboot 0 from the Serial console immediately instead of letting the board boot up. Nothing shows up.

I tried using Windows 10 and WSL yet nothing is showing up.

I just want to either fix the ssh connection issue or simply factory reset the entire thing.

I am using the Coral devboard and the Jetson T2 devboard. In order to send a model to them, the model has to have the extension .pb

Is there a link where the models already have a .pb extension? Currently I am using this link: TF_slim

All the models have the extension .ckpt and that is all. There is no .meta, or anything else. I do not know how to convert to .pb.

I am working in Colab. This is my code:

I cannot connect to my coral devboard via mdt shell over OTG USB-C port after following along here. Initially, I had the mendel 'public device cowardly refusal' issue as outlined in an q&answer here. I used serial console and nmtui to get the devboard on my LAN and I fixed that checking code on my host to match my LAN arrangement (172 instead of 192 block) and that worked to start connection via mdt shell.

Now, when connecting I get the below error about only allowing a single ssh key. I have ONLY connected via screen in serial console since I initially flashed the board. No one has ever ssh'd.

I have used "paramiko" to connect from my PC to a devboard, and execute a script. Then I am saving the results of this script in a list (output). I want to extract some values of the list and insert them into Elasticsearch. I have done it manually with the first result of the list. But how can I automate for the rest of the values? Do I need "regex"? Please give me some clues.

Thank you

THIS IS PART OF THE CODE THAT CONNECTS TO THE DEVBOARD, EXECUTES A SCRIPT AND RETRIEVES A LIST=output

I have a devboard (Google Coral). There are some scripts in there to perform image classification and object detection.

I want the results of those scripts (model, inference_time, score, image, etc) to be sent/indexed into Elasticsearch.

Also, there is a camera (came with the board) that I can use and take photos (there is a script for it too).

My approach so far has been:

1. I wrote a script using Paramiko to connect to the devboard. (It is in my PC)

2. Still, using Paramiko, I executed the scripts which are inside the devboard (turn on camera, wait until spacebar to take photo, make classification, send results to a server)

3. The results are indexed into Elasticsearch (it is installed in my PC).

I would like to know if there is a better way to do what I have done. What confuses me is that I have to connect to the board to execute a script (takes photo and performs classification). But I also have other script (in my PC) which connects to the devboard to give instructions, collect results and index them into Elasticsearch. Should I have put everything inside the devboard? Should I have used a serial communication? Should I have used subprocesses?

@JGK

This question is related to this one: How to use sockets to send user and password to a devboard using ssh

How can I put CODE A into a function? Explain me what am I doing wrong.

CODE A

This question is related to this other one: How to use sockets to send user and password to a devboard using ssh

I want to connect to the devboard in order to execute a script. All the outputs of that script I want to send to a Elasticsearch machine.

I can connect to the devboard (see IMAGE below) using my laptop which happens to have Elasticsearch installed. But, when I want to send data to the devboard, the script shows nothing. What I am doing is:

• As soon as you find mendel@undefined-eft:~$ , send the command: cd coral/tflite/python/examples/classification/Auto_benchmark\n

What am I doing wrong?

I am using sockets to connect to Coral devboard. I want to connect to the devboard and execute a classification script. Whenever I execute my code (lines below) , it shows "In [3]:" for around 10 seconds as if waiting for something. Then "In [4]" appears. What is it happening? I already sent the user and password.

This is my code: