DeepNAG: Deep Non-Adversarial Gesture Generation

TL;DR

This paper introduces DeepNAG, a novel gesture synthesis method that trains a generator without a discriminator, achieving faster training and higher realism than traditional GANs, and improves gesture recognition through data augmentation.

Contribution

The paper presents DeepNAG, a new differentiable loss-based approach for gesture synthesis that eliminates the need for a discriminator, enabling faster and more realistic data generation.

Findings

DeepNAG outperforms DeepGAN in accuracy and realism.

DeepNAG reduces training time by up to 17 times.

Synthesized gestures improve recognition performance across datasets.

Abstract

Synthetic data generation to improve classification performance (data augmentation) is a well-studied problem. Recently, generative adversarial networks (GAN) have shown superior image data augmentation performance, but their suitability in gesture synthesis has received inadequate attention. Further, GANs prohibitively require simultaneous generator and discriminator network training. We tackle both issues in this work. We first discuss a novel, device-agnostic GAN model for gesture synthesis called DeepGAN. Thereafter, we formulate DeepNAG by introducing a new differentiable loss function based on dynamic time warping and the average Hausdorff distance, which allows us to train DeepGAN's generator without requiring a discriminator. Through evaluations, we compare the utility of DeepGAN and DeepNAG against two alternative techniques for training five recognizers using data augmentation over six datasets. We further investigate the perceived quality of synthesized samples via an Amazon Mechanical Turk user study based on the HYPE benchmark. We find that DeepNAG outperforms DeepGAN in accuracy, training time (up to 17x faster), and realism, thereby opening the door to a new line of research in generator network design and training for gesture synthesis. Our source code is available at https://www.deepnag.com.