Skookum's Facebot?
The beginnings of a web app that can recognize Skookumites. Currently, there is a lot of manual setup.
Running
You must have the Facebox docker image running locally. TO do that, you'll need to get a key from machinebox.io which is FREE:
MB_KEY="YOUR MB key" docker run -p 8080:8080 -e "MB_KEY=$MB_KEY" machinebox/facebox
Then you need to train the Facebox with pictures (I was using the Bamboo HR pics)
and assign it a Name. The Name is used to lookup the sound file. You can upload photos
to the facebox by going to http://localhost:8080 and clicking on Try it now it the
"Teach faces" section. That will show a form that has a ID, Name, and URL field.
Put the URL to the image (you can paste in the URL to the Bamboo image) and a name and
click Post.
The /sound directory has the mp3s of the sounds. Each sound needs to be the
lowercase version of the name.mp3. I used espeakbox to synthesize text-to-speech.
Espeakbox offers another docker image, so:
docker run --name espeakbox -d -p 8082:8080 parente/espeakbox
curl http://localhost:8082/speech\?text\=Hello%20Enrique > sound/enrique.mp3
Once that's done, you can run this app, which starts a web server:
> go run main.go (or build it and run it)
Then, goto http://localhost:8081/static and put your face in the video window. It
should say "Hello "
Roadmap
- Get a real UI. Maybe an Avatar?
- Use Bamboo API to check employee details. Birthday, etc.
- Use Google AI API to be more interactive?
- Allow unknown user to have pic taken, register, etc.
- Keep interaction state. Meaning, already greeted, etc.
- Host facebox somewhere cloudy.
Questions
- Could we put this all over the office?
- How to incorporate listening? Like this