Page 54 - AI for Good Innovate for Impact
P. 54

AI for Good Innovate for Impact



                          Use Case 2: AI based Virtual Yoga Instructor for Good health












                      Country: India

                      Organization: Centre for Development of Advanced Computing, Pune

                      Contact Person(s):

                           Lakshmi Sandip Panat, lakshmip@ cdac .in
                           Swapna Yenishetti, swapnae@ cdac .in
                           Ganesh Karajkhede, ganeshk@ cdac .in


                      1      Use Case Summary Table


                       Item                      Details
                       Category                  Healthcare

                       Problem Addressed         Yoga learners and practitioners often face challenges of finding
                                                 convenient time slots for yoga practice, travel time required to
                                                 reach the yoga studio, personalized yoga coaching, tracking the
                                                 improvement in performing the yoga pose, affordability of yoga
                                                 classes, and customized content for specific health conditions like
                                                 yoga for diabetes, yoga for heart disease, etc. Also, the cost of
                                                 performing yoga poses incorrectly can result in negative effects
                                                 on the body, leading to muscle injury, joint pain, back pain, nerve
                                                 compression, pressure on internal organs, leading to injury, etc.
                                                 Yoga learners will benefit
                       Key Aspects of Solution   •  Real-time Yoga Pose Correction: Develop an AI algorithm to
                                                    provide immediate feedback on yoga postures during prac-
                                                    tice.
                                                 •  Improved Health Outcomes: Enable practitioners to achieve
                                                    better health benefits by ensuring correct and safe pose
                                                    execution.
                                                 •  Reduced Reliance on Human Instructors: Minimize the neces-
                                                    sity for constant human supervision, making yoga practice
                                                    more accessible.
                                                 •  Dynamic and Lightweight Processing: Ensure the system
                                                    can handle the dynamic nature of yoga poses with real-
                                                    time processing capabilities while remaining efficient and
                                                    resource-friendly.

                       Technology Keywords       Yoga Asanas, Yoga Pose correction, Convolutional Neural
                                                 Network (CNN), Movenet, TensorFlow/Keras [5]

                       Data Availability         Yoga data publicly available [1]

                       Metadata (Type of Data)   Images, Landmarks and Video Data






                  18
   49   50   51   52   53   54   55   56   57   58   59