Slide Extraction

Extract slides out of videos

Setup

You need to have ffmpeg installed & in path. I've noticed some inconsistencies between ffmpeg versions. Please use v3.0

Build a docker container for tensorflow server using the provided Dockerfile

cd docker
docker build --pull -t $USER/slide_detection_server -f ./Dockerfile.devel .

Install the required python dependencies

# Create a new env using virtualenv
virtualenv venv
source venv/bin/activate
# install required python packages
pip install -r requirements.txt

Extract slides from video

Start the container for tensorflow you have created

# attach the project dir as a volume to the container
docker run -p 9000:9000 -v PROJECT_DIR:/app $USER/slide_detection_server tensorflow_model_server --port=9000 --model_name=inception --model_base_path=/app/checkpoints

# just run this script and pass it the video url
# the detected slides are stored in the detected_slides folder
python detect.py --url URL_TO_YOUTUBE_VIDEO

Training the network

Import some data to train the network.

# The script downloads a bunch of youtube videos & serialzes them into bin
# vidoes stored in /data and training data in /images
# modify this script if you want to change data used to train the network
python import_data.py

Train the network

python image_retraining/retrain.py --image_dir images

Visualize

TODO

Architecture & memory constraints

TODO

Workflow to add more data

TODO