uMoDT: an unobtrusive multi-occupant detection and tracking using robust Kalman filter for real-time activity recognitio
- PDF / 1,848,304 Bytes
- 17 Pages / 595.276 x 790.866 pts Page_size
- 2 Downloads / 171 Views
REGULAR PAPER
uMoDT: an unobtrusive multi‑occupant detection and tracking using robust Kalman filter for real‑time activity recognition Muhammad Asif Razzaq1 · Javier Medina Quero2 · Ian Cleland3 · Chris Nugent3 · Usman Akhtar1 · Hafiz Syed Muhammad Bilal1 · Ubaid Ur Rehman1 · Sungyoung Lee1 Received: 9 August 2019 / Accepted: 9 June 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Human activity recognition (HAR) is an important branch of human-centered research. Advances in wearable and unobtrusive technologies offer many opportunities for HAR. While much progress has been made in HAR using wearable technology, it still remains a challenging task using unobtrusive (non-wearable) sensors. This paper investigates detection and tracking of multi-occupant HAR in a smart-home environment, using a novel low-resolution Thermal Vision Sensor (TVS). Specifically, the research presents the development and implementation of a two-step framework, consisting of a Computer Vision-based method to detect and track multiple occupants combined with Convolutional Neural Network (CNN)-based HAR. The proposed algorithm uses frame difference over consecutive frames for occupant detection, a set of morphological operations to refine identified objects, and features are extracted before applying a Kalman filter for tracking. Laterally, a 19-layer CNN architecture is used for HAR and afterward the results from both methods are fused using time interval-based sliding window. This approach is evaluated through a series of experiments based on benchmark Thermal Infrared datasets (VOT-TIR2016) and multi-occupant data collected from TVS. Results demonstrate that the proposed framework is capable of detecting and tracking 88.46% of multi-occupants with a classification accuracy of 90.99% for HAR. Keywords Human activity recognition · Image processing · Object detection · Tracking · Classification
1 Introduction Over several decades, advances in pervasive computing have offered great promise towards the potential of indoor localization and Human Activity Recognition (HAR) [1]. Over this period, significant research effort has been targeted Communicated by C. Xu.
towards the creation of solutions that can reliably monitor individuals through the use of on-body wearable sensors, dense sensors, and vision sensors [2]. Whilst results utilizing on-body sensors have improved greatly, wearable solutions are often said to be impractical, as they can be difficult to carry or inconvenient to wear continuously [3]. Additionally, vision sensors capable of capturing RGB or grayscale images have been studied intensively within the Computer
* Sungyoung Lee [email protected]
Hafiz Syed Muhammad Bilal [email protected]
Muhammad Asif Razzaq [email protected]
Ubaid Ur Rehman [email protected]
Javier Medina Quero [email protected]
1
Ian Cleland [email protected]
Ubiquitous Computing Lab, Department of Computer Engineering, Kyung Hee University, Seocheon‑dong, Giheung‑gu, Y
Data Loading...
