A Note on the Speed-Accuracy Trade-off of Super-Resolution for Human Pose Estimation

View at publisher Summarise with VIRAT
Authors: Aki Shigesawa, Chinami Fukui, Masahiro Yagi, Sho Takahashi, Toshio Yoshii, Marco Bertini
Cited by: 0
References: 7

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This work evaluates the trade-off between speed and accuracy when using super-resolution as a preprocessing step for human pose estimation on low-resolution images. We assess pose estimation accuracy and inference speed for various super-resolution methods on a public dataset. Results show that learning-based super-resolution methods significantly outperform the Bicubic interpolation baseline. Furthermore, we quantify a clear trade-off between the higher-accuracy SRGAN and the faster FSRUNet. This analysis serves as a practical guide for selecting an appropriate model based on application requirements.

Fields of Study & Subjects

Advanced Image Processing Techniques Human Pose and Action Recognition Advanced Neural Network Applications

Publication details

Journal / source
International Conference on Consumer Electronics
Pages
1–3
Publication date
3 February 2026
Language
EN