Dense CNN and IndRNN for the Sussex-Huawei Locomotion-Transportation Recognition Challenge
UbiComp/ISWC 2021 - Adjunct Proceedings of the 2021 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2021 ACM International Symposium on Wearable Computers
The Sussex-Huawei Locomotion Challenge (SHL) 2021 organized at the HASCA Workshop of UbiComp 2021 is to recognize human activity by using GPS reception, GPS location, Wifi reception and GSM cell tower scans data. Compared with the previous challenge, this challenge is more difficult. In this paper, our team (NUC) summarize our submission to the competition. We propose a framework of deep learning including one-dimensional (1D) Dense CNN and Dense IndRNN networks to explore short and long-Term spatial temporal information. The location and Wifi features are extracted and normalized to feed into deep learning networks. Finally, the decision fusion is utilized to improve performance.
Open Access Status
This publication is not available as open access
National Natural Science Foundation of China