TY - JOUR
T1 - Rank pooling approach for wearable sensor-based adls recognition
AU - Nisar, Muhammad Adeel
AU - Shirahama, Kimiaki
AU - Li, Frédéric
AU - Huang, Xinyu
AU - Grzegorzek, Marcin
N1 - Funding Information:
We use rank pooling for sensor based data to recognize 7 composite activities. We produce a large dataset that contains 9029 examples of 61 atomic activities and 890 examples of 7 composite activities. This dataset is collected within the Cognitive Village Project supported by the German Federal Ministry of Education and Research (BMBF), and named CogAge dataset for the sake of simplicity.
Funding Information:
Funding: Research and development activities leading to this article have been supported by the German Research Foundation (DFG) as part of the research training group GRK 1564 “Imaging New Modalities”, and the German Federal Ministry of Education and Research (BMBF) within the project Cognitive Village (grant number: 16SV7223K http://cognitive-village.fokos.org/).
Publisher Copyright:
© 2020 by the authors. Licensee MDPI, Basel, Switzerland.
Copyright:
Copyright 2020 Elsevier B.V., All rights reserved.
PY - 2020/6
Y1 - 2020/6
N2 - This paper addresses wearable-based recognition of Activities of Daily Living (ADLs) which are composed of several repetitive and concurrent short movements having temporal dependencies. It is improbable to directly use sensor data to recognize these long-term composite activities because two examples (data sequences) of the same ADL result in largely diverse sensory data. However, they may be similar in terms of more semantic and meaningful short-term atomic actions. Therefore, we propose a two-level hierarchical model for recognition of ADLs. Firstly, atomic activities are detected and their probabilistic scores are generated at the lower level. Secondly, we deal with the temporal transitions of atomic activities using a temporal pooling method, rank pooling. This enables us to encode the ordering of probabilistic scores for atomic activities at the higher level of our model. Rank pooling leads to a 5–13% improvement in results as compared to the other popularly used techniques. We also produce a large dataset of 61 atomic and 7 composite activities for our experiments.
AB - This paper addresses wearable-based recognition of Activities of Daily Living (ADLs) which are composed of several repetitive and concurrent short movements having temporal dependencies. It is improbable to directly use sensor data to recognize these long-term composite activities because two examples (data sequences) of the same ADL result in largely diverse sensory data. However, they may be similar in terms of more semantic and meaningful short-term atomic actions. Therefore, we propose a two-level hierarchical model for recognition of ADLs. Firstly, atomic activities are detected and their probabilistic scores are generated at the lower level. Secondly, we deal with the temporal transitions of atomic activities using a temporal pooling method, rank pooling. This enables us to encode the ordering of probabilistic scores for atomic activities at the higher level of our model. Rank pooling leads to a 5–13% improvement in results as compared to the other popularly used techniques. We also produce a large dataset of 61 atomic and 7 composite activities for our experiments.
UR - http://www.scopus.com/inward/record.url?scp=85087014660&partnerID=8YFLogxK
U2 - 10.3390/s20123463
DO - 10.3390/s20123463
M3 - Journal articles
C2 - 32575451
AN - SCOPUS:85087014660
SN - 1424-8220
VL - 20
SP - 1
EP - 21
JO - Sensors (Switzerland)
JF - Sensors (Switzerland)
IS - 12
M1 - 3463
ER -
