Sign Language Production Using Generative AI
D. Karthika Renuka, L. Ashok Kumar, K R Harini, Lonka Nithin, P R Rishi Khanna, J Suruthika · 2024
Sign language is vital for communication among the deaf and hard of hearing, but a lack of interpreters limits accessibility. This research utilizes Generative Adversarial Networks (GANs), specifically a Conditional GAN (CGAN) based on the pix2pix model, to generate sign language gestures from skeletal pose data. Keypoints extracted through OpenPose are converted into realistic body movements, with a focus on fluidity. While the CGAN successfully generates natural movements, replicating detailed hand gestures remains challenging. Future work will aim to improve hand gesture accuracy and integrate real-time speech recognition to enhance accessibility.