WKT05a-21 Exercise Guidance System

Subscribe
Notify of
9 Comments
Inline Feedbacks
View all comments
CAO, Xuanyu(@eexcao)
January 19, 2022 4:05 pm

Timely project. What image processing/ computer vision methods did you use to estimate the posture and guide the exercise?

Last edited 10 months ago by CAO, Xuanyu
HUI, Chi To(@cthui)
January 19, 2022 4:30 pm
Reply to  CAO, Xuanyu

Thank you for your comments. We use Opencv and mediapipe in python. Building the body landmarks to track the movement. We can obtain the data of the landmarks. At present, we plan to calculate the angles of the three landmarks to determine whether the user’s actions meet the standard.

In the other part, we intend to train a model to rate the user’s actions and give appropriate suggestions to improve the user’s training performance.

LAI, Chi Wan(@cwlaiaf)
January 19, 2022 3:08 pm

Cool project! Will the guidance system work in real-time response?

CHAN, Chun Yeung(@cychancu)
January 19, 2022 3:18 pm
Reply to  LAI, Chi Wan

Thank you for your comments, our system will act as an expert trainer for guiding the user’s posture in real-time 

HUI, Chi To(@cthui)
January 19, 2022 3:21 pm
Reply to  LAI, Chi Wan

Thank you for your question. The system can estimate users pose and track movement in real-time. However, we are still working on ways to train a model to grade user performance and provide appropriate feedback.

YEUNG, Chun Kit(@ckyeungaj)
January 19, 2022 2:53 pm

Great job! I saw you have two technical challenges, how would you solve them?

LAU, Sung Ming(@smlauac)
January 19, 2022 3:31 pm

Thank you for your comment. We have tested 540p and 720p videos respectively. In the case of the same video and different resolutions, the accuracy and fps of the two are not much different, so we may use different video resolutions according to the complexity of the action, such as a lower resolution for simple poses.

LI, Chun Ming(@cmliag)
January 19, 2022 2:39 pm

Great work in the poster design and the descriptions in each sub-heading! The overview part is convincing to me. Such a fitness guidance system product definitely needs to come to the market asap. By the way, my group are also doing a similar project, but our system is more mobile orientated. Would you mind checking out our poster and sharing your opinions? 🙂

CHAN, Chun Yeung(@cychancu)
January 19, 2022 2:54 pm
Reply to  LI, Chun Ming

Thank you for your comments, I will check out your poster now.