Face tracking and follow using Google USB TPU

Introduction

This is the step by step tutorial for making simple follow me demo on Ohmni with Google Edge TPU device. The face tracking model and following code is provided in a docker container that you build and run on Ohmni. 

What you need

Hardware

1. Ohmni Robot
2. Coral USB accelerator (https://coral.ai/products/accelerator/)
3. Ohmni USB expansion kit (https://store.ohmnilabs.com/collections/developer-edition-expansion-kits/products/usb-expansion-kit)

Software

1. Face tracking model and following code (https://gitlab.com/ohmni-sdk/demo-tpufacetracking) (Please reach out to [e-mail] referencing this guide to be added to access).
2. Ohmni TPU docker (link to docker file and install script)

Install TPU

1. Plug the Coral USB accelerator into any ports on the Ohmni USB extension kit.

Install Ohmni TPU Docker

* skip this section if you already have an existing Ohmni TPU docker image

1. Remote access the robot using SSH or ADB
2. Create /var/dockerhome if it doesn't exist and cd to it.
3. Copy dockerfile and install.sh to /var/dockerhome
4. From /var/dockerhome build the docker image,
   ohmni_up:/# docker build --network="host" --tag ohmnilabs/tpusupportdocker .

Install Face Tracking Docker

1. Copy the object detection sample code to a new folder called demo-tpufacetracking  in your working directory
2. Copy the dockerfile to the working directory
3. From the working directory, build the docker image,
   ohmni_up:/# docker build --network="host" --tag ohmnilabs/tpusupportdocker:facefollowdemo .
    
   

Run the code

Note: ohmni.sh will grab the image stream from the camera, so you will need to run any app that will start the camera stream first. In this case, we will run the Android default camera app.

ohmni_up:/# monkey -p net.sourceforge.opencamera -c android.intent.category.LAUNCHER 1

ohmni_up:/# docker-ohmnirun ohmnilabs/tpusupportdocker:facefollowdemo