# Harnessing GANs for Zero-shot Learning of New Classes in Visual Speech Recognition

## TL;DR

This paper introduces a GAN-based zero-shot learning approach for visual speech recognition, significantly improving accuracy on unseen classes and demonstrating language independence, which is a novel contribution in the VSR domain.

## Contribution

The paper presents the first empirical use of GANs to generate training samples for unseen classes in VSR, enabling zero-shot learning and language-agnostic video generation.

## Key findings

- Increased VSR accuracy by 27% with generated unseen class samples.
- Models are capable of generating videos for new languages like Hindi.
- First empirical evidence of GANs aiding zero-shot learning in VSR.

## Abstract

Visual Speech Recognition (VSR) is the process of recognizing or interpreting speech by watching the lip movements of the speaker. Recent machine learning based approaches model VSR as a classification problem; however, the scarcity of training data leads to error-prone systems with very low accuracies in predicting unseen classes. To solve this problem, we present a novel approach to zero-shot learning by generating new classes using Generative Adversarial Networks (GANs), and show how the addition of unseen class samples increases the accuracy of a VSR system by a significant margin of 27% and allows it to handle speaker-independent out-of-vocabulary phrases. We also show that our models are language agnostic and therefore capable of seamlessly generating, using English training data, videos for a new language (Hindi). To the best of our knowledge, this is the first work to show empirical evidence of the use of GANs for generating training samples of unseen classes in the domain of VSR, hence facilitating zero-shot learning. We make the added videos for new classes publicly available along with our code.