Scholarlink

Recently Generative Adversarial Networks or GANs have gained a lot of popularity, causing them to be adopted in a variety of fields where rich and real-time data need to be created. We examine in this paper the potential for employing GANs for bridging communication disparities between deaf individuals and hearing populations within the realm of American Sign Language production. A new framework is proposed by us to create expressive and precise ASL gestures which can be utilized for real-time text-to-sign translation. By and large, this study promotes the development of aids for the hearing-impaired people, therefore underscoring the significant role of GANs in social inclusion and general communication for all groups. Our paper analyzes the ways in which GAN architectures may be devised with respect to cochlear implants (CIs) requirements within ASL context. We have looked at 3 distinct flavors of GANs in this undertaking. First, we tried a simple GAN architecture before moving to Deep Convolutional GANs (DCGANs). Super Resolution GANs (SRGANs) on the other hand produced good results with optimal Generator and Discriminator losses. Through experimentation, we argue that our suggested technique is proficient and versatile for generating correct ASL images reflecting the intended language. GAN-based ASL generation systems have a range of uses that could help people who are deaf or have hearing problems become better educated, find jobs more easily or socialize more effectively.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Harnessing the Power of Generative Adversarial Networks for American Sign Language Generation

  • 章节

  • 发表日期：

    2026年

查看全文
  • Shreya Mahajan,
  • Ayushi Patil,
  • Jinal Menpara,
  • K. Nandhini,
  • Shilpa Gite,
  • Biswajeet Pradhan

摘要

Recently Generative Adversarial Networks or GANs have gained a lot of popularity, causing them to be adopted in a variety of fields where rich and real-time data need to be created. We examine in this paper the potential for employing GANs for bridging communication disparities between deaf individuals and hearing populations within the realm of American Sign Language production. A new framework is proposed by us to create expressive and precise ASL gestures which can be utilized for real-time text-to-sign translation. By and large, this study promotes the development of aids for the hearing-impaired people, therefore underscoring the significant role of GANs in social inclusion and general communication for all groups. Our paper analyzes the ways in which GAN architectures may be devised with respect to cochlear implants (CIs) requirements within ASL context. We have looked at 3 distinct flavors of GANs in this undertaking. First, we tried a simple GAN architecture before moving to Deep Convolutional GANs (DCGANs). Super Resolution GANs (SRGANs) on the other hand produced good results with optimal Generator and Discriminator losses. Through experimentation, we argue that our suggested technique is proficient and versatile for generating correct ASL images reflecting the intended language. GAN-based ASL generation systems have a range of uses that could help people who are deaf or have hearing problems become better educated, find jobs more easily or socialize more effectively.