A Web-Based Fitness Assistant: Implementing Pose Estimation with OpenCV and MediaPipe

In this paper, we addressed the problem of real-time posture detection and exercise monitoring in web-based fitness systems. We proposed an improved fitness assistant framework, FitPose v1, based on MediaPipe and OpenCV technologies. PAT, MSC, and FOM modules are introduced in the framework to improve body posture understanding and sensitivity to different exercise movements more effectively. Furthermore, compared with traditional posture tracking methods, our framework achieves better posture detection accuracy while still maintaining real-time performance with lower computational complexity. As a result, it is more suitable for deployment in intelligent fitness monitoring systems where devices often have limited computational power and memory resources. Therefore, the proposed framework, FitPose v1, can serve as an efficient and reliable solution for real-time fitness assistance with high accuracy and smooth performance.

1)  Sutton, R. S., and Barto, A. G., “Reinforcement Learning: An Introduction,” 2nd ed., Cambridge, MA, USA: MIT Press, 2018. 2)  Lee, H. L., Padmanabhan, V., and Whang, S., “The Bullwhip Effect in Supply Chains,” Management Science, vol. 43, no. 4, pp. 546–558, 1997. 3)  Simchi-Levi, D., Kaminsky, P., and Simchi-Levi, E., “Designing and Managing the Supply Chain: Concepts, Strategies, and Case Studies,” 3rd ed., New York, NY, USA: McGraw-Hill, 2007. 4)  Schulman, J., Wolski, F., Dhariwal, P., Radford, A., and Klimov, O., “Proximal Policy Optimization Algorithms,” arXiv preprint arXiv:1707.06347, 2017. 5)  Mnih, V., Kavukcuoglu, K., Silver, D., et al., “Human-Level Control Through Deep Reinforcement Learning,” Nature, vol. 518, pp. 529–533, 2015. 6)  Silver, D., Huang, A., Maddison, C. J., et al., “Mastering the Game of Go with Deep Neural Networks and Tree Search,” Nature, vol. 529, pp. 484–489, 2016. 7)  Lowe, R., Wu, Y., Tamar, A., et al., “Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments,” in Advances in Neural Information Processing