Towards Practical Lipreading with Distilled and Efficient Models

TL;DR

This paper advances lipreading technology by achieving higher accuracy with self-distillation, introduces a computationally efficient model architecture, and demonstrates lightweight models suitable for practical deployment.

Contribution

It presents a novel depthwise separable TCN architecture and uses self-distillation to create lightweight models with competitive accuracy and significantly reduced computational costs.

Findings

State-of-the-art accuracy on LRW and LRW-1000 datasets.

Lightweight models with 8.2x and 3.9x fewer parameters.

Models suitable for real-world deployment.

Abstract

Lipreading has witnessed a lot of progress due to the resurgence of neural networks. Recent works have placed emphasis on aspects such as improving performance by finding the optimal architecture or improving generalization. However, there is still a significant gap between the current methodologies and the requirements for an effective deployment of lipreading in practical scenarios. In this work, we propose a series of innovations that significantly bridge that gap: first, we raise the state-of-the-art performance by a wide margin on LRW and LRW-1000 to 88.5% and 46.6%, respectively using self-distillation. Secondly, we propose a series of architectural changes, including a novel Depthwise Separable Temporal Convolutional Network (DS-TCN) head, that slashes the computational cost to a fraction of the (already quite efficient) original model. Thirdly, we show that knowledge distillation is a very effective tool for recovering performance of the lightweight models. This results in a range of models with different accuracy-efficiency trade-offs. However, our most promising lightweight models are on par with the current state-of-the-art while showing a reduction of 8.2x and 3.9x in terms of computational cost and number of parameters, respectively, which we hope will enable the deployment of lipreading models in practical applications.