面向头部姿态估计多输出回归模型增强型边缘推理的参数选择比较研究
计算机视觉与模式识别
2023-02-02 v1 人工智能
摘要
基于幅度的剪枝是一种用于优化深度学习模型以实现边缘推理的技术。我们已实现超过 75% 的模型尺寸缩减，且精度高于原始的头部姿态估计多输出回归模型。
引用
@article{arxiv.2302.00592,
title = {Comparative Study of Parameter Selection for Enhanced Edge Inference for a Multi-Output Regression model for Head Pose Estimation},
author = {Asiri Lindamulage and Nuwan Kodagoda and Shyam Reyal and Pradeepa Samarasinghe and Pratheepan Yogarajah},
journal= {arXiv preprint arXiv:2302.00592},
year = {2023}
}
备注
Conference:- in TENCON 2022 - 2022 IEEE Region 10 Conference (TENCON)