Computing FSK using Python and OpenNSFW

Courtesy of Yahoo engineering:

https://github.com/yahoo/open_nsfw

Detailed explanation: https://yahooeng.tumblr.com/post/151148689421/open-sourcing-a-deep-learning-solution-for

Installing the stuff we need

Start docker-machine (if Mac)
docker build -t caffe:cpu https://raw.githubusercontent.com/BVLC/caffe/master/docker/standalone/cpu/Dockerfile
docker run caffe:cpu caffe --version
git clone https://github.com/yahoo/open_nsfw.git
cd open_nsfw
docker run --volume=$(pwd):/workspace caffe:cpu
python ./classify_nsfw.py --model_def nsfw_model/deploy.prototxt --pretrained_model nsfw_model/resnet_50_1by2_nsfw.caffemodel test_image.jpg

You will get a result: NSFW score: 0.14057905972

Python Libraries

virtualenv venv
source venv/bin/activate
pip install moviepy
pip install youtube_dl
pip install shlex

Now we are all set

Get a test video using youtube_dl

youtube-dl --merge-output-format mp4 https://www.youtube.com/watch?v=P8IhIPq42Wg

Work on the video and split it into frames

See splitandtest.py

Send every frame into classification

See classify.py

Now automate it and plot and do more