TY - JOUR
AU - Govindaraj, Vishnuvarthanan
AB - The research presents an interactive toy agent leveraging a deep learning approach for treating autistic kids by making them learn Yoga. The objective of the proposed toy is to understand the basic needs of autistic kids and make them socially adaptable to their surrounding environment. Since kids with autism face social insecurities while interacting and communicating with people, we introduce an interactive toy to accompany the kid, making him or her more likely to act as a companion. The toy is orchestrated with IoT and the Deep Learning framework (HARNet) which makes it interactively instruct Yoga Asana to the autistic kid. The motion of the toy is controlled by touch sensors, and interaction is developed through the recognition of Yogo postures performed by the kid. This paper uses snippets of data in the Yoga-82 dataset. The gestures of Yoga asanas are leveraged, and the same is used for modeling HARNet. Empirical evaluations show that HARNet exhibits an accuracy of 98.52% against the Yoga-82 dataset. The cost of the Toy framework is also compared with state-of-the-art research on Humanoid Toys and the economic range of the proposed framework is evident.
TI - HARNet: design and evaluation of a deep genetic algorithm for recognizing yoga postures
JF - Signal Image and Video Processing
DO - 10.1007/s11760-024-03173-6
DA - 2024-08-01
UR - https://www.deepdyve.com/lp/springer-journals/harnet-design-and-evaluation-of-a-deep-genetic-algorithm-for-T0YNdlvjKT
SP - 553
EP - 564
VL - 18
IS - Suppl 1
DP - DeepDyve
ER -