Data Preprocessing for Oxford's LRW Youtube Dataset
This project was done as part of the final project for the CIE 555: Neural Networks and Deep Learning Course at Zewail City, taught by Dr. Mohammed Elshennawy.
The notebook contains fully automated video preprocessing for the dataset from OXFORD Visual Geometry Group's dataset (LRW: Lip Reading in the Wild) by BBC.
This is not by any means an official implementation.
The preprocessing was done using ffmpeg, youtube-dl, and Python.
The notebook assumes running in a google colab environment. If not, please remove the first part (mounting drive, and unzipping) and edit the following code accordingly.