DEPRECATED: MoodyKubie

MoodyKubie is a kubernetes-native face recognition service built to run on top of an ARM clusters of Cubieboard 2.0 like the Cubienetes.

It was built as a demo for KubeCon Austin 2017 by a team at Weaveworks.

Images

The recognition service image can be found on Docker Hub as weaveworks/moodykubie-service and will expose the recognition service on port 8989.

The UI image can be found on Docker Hub as weaveworks/moodykubie-ui and will expose the UI on port 9000.

Developing

You don't need to build the Docker images to work on the MoodyKubie.

You'll need Node and Go installed. Then, in one terminal run the recognition service using the following commands:

# Navigate to the backend service app
➤ cd moodykubie-service/app/

# Download and install dependencies
➤ npm install
...

# Run the application
➤ node index.js
Emotion detection service live on port 8989

In another terminal, run the UI:

# Navigate to the UI
➤ cd moodykubie-ui

# Run the app
➤ go run main.go
2017/11/30 19:21:51 Starting HTTP server on port 9000

The UI will be served on port 9000. navigate to http://localhost:9000 to view it. The UI client files can be found in moodykubie-ui/static

Run

Docker

docker run --rm --name moodykubie-service \
    -p 8989:8989 \
    weaveworks/moodykubie-service:latest

docker run --rm --name moodykubie-ui \
    --link moodykubie-service \
    -e SERVICE_HOST=moodykubie-service \
    -e SERVICE_PORT=8989 \
    -p 9000:9000 \
    weaveworks/moodykubie-ui:latest

### Manually trigger a face recognition from web browser.

curl localhost:8989/metrics
# HELP emotions_total Total number of recognised emotions.
# TYPE emotions_total counter
emotions_total{emotion="anger"} 1

Kubernetes

cd /path/to/this/project
kubectl apply -f kubernetes

Build & Push

Get hold of an ARM 32 bit machine (Cubieboard, RPi, etc.) or start a C1 instance on Scaleway, and run the below:

apt-get remove docker docker-engine docker.io
apt-get -yq update
apt-get -yq install \
    apt-transport-https \
    ca-certificates \
    curl \
    git \
    software-properties-common
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
apt-key fingerprint 0EBFCD88
dpkg --print-architecture
add-apt-repository \
    "deb [arch=armhf] https://download.docker.com/linux/ubuntu \
    $(lsb_release -cs) \
    stable"
apt-get -yq update
apt-get -yq install docker-ce
curl -fsSLO https://storage.googleapis.com/golang/go1.9.2.linux-armv6l.tar.gz && \
    tar -C /usr/local -xzf go1.9.2.linux-armv6l.tar.gz && \
    rm -f go1.9.2.linux-armv6l.tar.gz
export PATH=$PATH:/usr/local/go/bin
docker login
git clone https://github.com/weaveworks/MoodyKubie.git
cd MoodyKubie
cd moodykubie-service
make all
cd ..
cd moodykubie-ui
make all

Case

The cut files are available in the Case directory. Note that when opening them in your favorite vector editor, the cut lines may not show up as they are very thin. You may have to turn on "outline" mode. In Inkscape, this can be found in View > Display Mode menu.

Getting Help

If you have any questions about, feedback for or problems with MoodyKubie:

• Invite yourself to the Weave Users Slack.
• Ask a question on the #general slack channel.
• File an issue.

Weaveworks follows the CNCF Code of Conduct. Instances of abusive, harassing, or otherwise unacceptable behavior may be reported by contacting a Weaveworks project maintainer, or Alexis Richardson (alexis@weave.works).

Your feedback is always welcome!