12th EAI International Conference on Mobile Multimedia Communications, Mobimedia 2019, 29th - 30th Jun 2019, Weihai, China

Research Article

Improved Capsule Network for Gaze Estimation in Wireless Sensor Networks

Download394 downloads
  • @INPROCEEDINGS{10.4108/eai.29-6-2019.2282839,
        author={Mingyuan Luo and Xi Liu and Wei Wang and Wei Huang},
        title={Improved Capsule Network for Gaze Estimation in Wireless Sensor Networks},
        proceedings={12th EAI International Conference on Mobile Multimedia Communications, Mobimedia 2019, 29th - 30th Jun 2019, Weihai, China},
        publisher={EAI},
        proceedings_a={MOBIMEDIA},
        year={2019},
        month={6},
        keywords={gaze estimation capsule network multi-layer capsule network},
        doi={10.4108/eai.29-6-2019.2282839}
    }
    
  • Mingyuan Luo
    Xi Liu
    Wei Wang
    Wei Huang
    Year: 2019
    Improved Capsule Network for Gaze Estimation in Wireless Sensor Networks
    MOBIMEDIA
    EAI
    DOI: 10.4108/eai.29-6-2019.2282839
Mingyuan Luo1, Xi Liu1, Wei Wang2, Wei Huang1,*
  • 1: Nanchang University
  • 2: Chang’an University
*Contact email: n060101@e.ntu.edu.sg

Abstract

In this study, aiming at the problem of gaze estimation in the wireless sensor network in the car, we use image-based method to estimate gaze based on the single camera sensor. We use the deep learning model and propose the improved model from three aspects based on the original capsule network. The first is to increase the convolution layer, the second is to increase the capsule layer, and the third is to widen the capsule layer in the network. Through many contrast experiments, it is proved that the appropriate use of the first or second improved method can achieve performance over other comparison models, and the prediction results of gaze estimation are almost no different from the real gaze direction.