Summary of Yoga Pose Classification Using Transfer Learning, by M. M. Akash et al.
Yoga Pose Classification Using Transfer Learning
by M. M. Akash, Rahul Deb Mohalder, Md. Al Mamun Khan, Laboni Paul, Ferdous Bin Ali
First submitted to arxiv on: 29 Oct 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This research paper presents a solution to the problem of human pose estimation in yoga, which has become increasingly important for maintaining physical and mental well-being. The authors address this challenge by developing a large-scale benchmark dataset called Yoga-82, comprising 82 classes with challenging positions that require precise annotations. To achieve better results, they fine-tune various pre-trained architectures, including VGG-16, ResNet-50, ResNet-101, and DenseNet-121. Additionally, the authors employ Neural Architecture Search to add more layers on top of these pre-trained models. The experimental results show that DenseNet-121 outperforms the current state-of-the-art result, achieving a top-1 accuracy of 85% and a top-5 accuracy of 96%. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Yoga is becoming super important for keeping our bodies and minds healthy. But it can be hard to fit in gym time when we’re working from home and life gets busy! One big problem is figuring out how to recognize different yoga poses, especially the tricky ones. The authors created a special dataset called Yoga-82 with 82 classes of challenging poses that are really hard to label precisely. They used four different models (VGG-16, ResNet-50, ResNet-101, and DenseNet-121) and tweaked them in different ways to get better results. They even used a cool tool called Neural Architecture Search to add more layers on top of these models. The results show that one model, DenseNet-121, did really well, achieving an accuracy of 85% for the most common poses and 96% for the top five. |
Keywords
» Artificial intelligence » Pose estimation » Resnet
