Computer vision project attempting to discover and track faces in unconstrained video. https://sites.google.com/vt.edu/adaptivefacetracking/home
As of right now face detection is the limiting factor
###Training Tf Models
cd tf/tensorflow/
source settings.sh
./run.sh
Check run.sh script for settings used.
###Running Face Tracking Make sure to place a relevant BigBangTheory clip in the root repo directory in .mp4 format. Set the video_path variable in detect_and_label_video.py
video_path = "./TBBT_S10E16.mp4"
Then run the script
./run_video.sh