Build docker image

docker build -t="anphunl/openface_demo" .

Start docker container

docker run -t -i -p 6006:6006 -p 5000:5000 -v ~/OpenfaceDemoData:/OpenfaceDemo/data anphunl/openface_demo

Prepare data

1. Download data at https://drive.google.com/file/d/146RG_go_yT0sr3Opn3DuD-DcPIgkIsY2/view?usp=sharing
2. Extract data to shared folder at ~/OpenfaceDemoData/raw on host machine.
3. Align face, run within container

for N in {1..4}; do /root/openface/util/align-dlib.py data/raw align outerEyesAndNose data/aligned --size 96 & done

3. Visualize data with Tensorboard:

python visualize.py

tensorboard --logdir=data/visuallize

4. Open browser with URL: http://localhost:6006 and see the visuallization.

Create server classify other images:

1. Create features:

/root/openface/batch-represent/main.lua -outDir data/model -data data/aligned/

2. Create classify model file:

/root/openface/demos/classifier.py train data/model/

3. Run server:

export FLASK_APP=server.py

flask run --host=0.0.0.0

4. Open browser with URL: http://localhost:5000 and check it.