Enhancing Yoga Pose Estimation Accuracy Using Optimized Mask R-CNN Model
Deepak Shukla, Maya Rathore · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2025
Yoga pose estimation is important for fitness, healthcare, and rehabilitation applications, existing models such as AlexNet, VGG, and ResNet cannot accurately recognize detailed key points or handle complex postures.To tackle these issues, this paper presents an improved mask R-CNN with better feature aggregation and segmentation and introduces a key point detection branch.Performance analysis demonstrates the effectiveness of our proposed model by improved values of mAP, [email protected], and [email protected] metrics.This approach has been experimentally shown to be used for realtime recovery from yoga poses.This work pushes forward the accuracy and scalability of pose estimation for widespread fitness and healthcare applications.