Abstract: As smartphones are continually upgrading, their software and hardware are getting smarter, so the smartphone-based human activity recognition will be described more refined, complex and detailed. In this context, we analyzed a set of experimental data, obtained by observing and measuring 30 volunteers with six activities of daily living (ADL). Due to the large sample size, especially a 561-feature vector with time and frequency domain variables, cleaning these intractable features and training a proper model become extremely challenging. After a series of feature selection and parameters adjustments, a well-performed SVM classifier has been trained.
Abstract: We propose to record Activities of Daily Living
(ADLs) of elderly people using a vision-based system so as to provide
better assistive and personalization technologies. Current ADL-related
research is based on data collected with help from non-elderly subjects
in laboratory environments and the activities performed are predetermined
for the sole purpose of data collection. To obtain more
realistic datasets for the application, we recorded ADLs for the elderly
with data collected from real-world environment involving real elderly
subjects. Motivated by the need to collect data for more effective
research related to elderly care, we chose to collect data in the room of
an elderly person. Specifically, we installed Kinect, a vision-based
sensor on the ceiling, to capture the activities that the elderly subject
performs in the morning every day. Based on the data, we identified
12 morning activities that the elderly person performs daily. To
recognize these activities, we created a HARELCARE framework to
investigate into the effectiveness of existing Human Activity
Recognition (HAR) algorithms and propose the use of a transfer
learning algorithm for HAR. We compared the performance, in terms
of accuracy, and training progress. Although the collected dataset is
relatively small, the proposed algorithm has a good potential to be
applied to all daily routine activities for healthcare purposes such as
evidence-based diagnosis and treatment.
Abstract: In this paper, we want to determine whether the accurate prediction of nicotine concentration can be obtained by using a network of smart objects and an e-cigarette. The approach consists of, first, the recognition of factors influencing smoking cessation such as physical activity recognition and participant’s behaviors (using both smartphone and smartwatch), then the prediction of the configuration of the e-cigarette (in terms of nicotine concentration, power, and resistance of e-cigarette). The study uses a network of commonly connected objects; a smartwatch, a smartphone, and an e-cigarette transported by the participants during an uncontrolled experiment. The data obtained from sensors carried in the three devices were trained by a Long short-term memory algorithm (LSTM). Results show that our LSTM-based model allows predicting the configuration of the e-cigarette in terms of nicotine concentration, power, and resistance with a root mean square error percentage of 12.9%, 9.15%, and 11.84%, respectively. This study can help to better control consumption of nicotine and offer an intelligent configuration of the e-cigarette to users.
Abstract: In this paper, we analyze major components of activity recognition (AR) in wearable device with 9-axis sensors and sensor fusion filters. 9-axis sensors commonly include 3-axis accelerometer, 3-axis gyroscope and 3-axis magnetometer. We chose sensor fusion filters as Kalman filter and Direction Cosine Matrix (DCM) filter. We also construct sensor fusion data from each activity sensor data and perform classification by accuracy of AR using Naïve Bayes and SVM. According to the classification results, we observed that the DCM filter and the specific combination of the sensing axes are more effective for AR in wearable devices while classifying walking, running, ascending and descending.
Abstract: As smartphones are equipped with various sensors,
there have been many studies focused on using these sensors to create
valuable applications. Human activity recognition is one such
application motivated by various welfare applications, such as the
support for the elderly, measurement of calorie consumption, lifestyle
and exercise patterns analyses, and so on. One of the challenges one
faces when using smartphone sensors for activity recognition is that
the number of sensors should be minimized to save battery power. In
this paper, we show that a fairly accurate classifier can be built that
can distinguish ten different activities by using only a single sensor
data, i.e., the smartphone accelerometer data. The approach that we
adopt to deal with this twelve-class problem uses various methods.
The features used for classifying these activities include not only the
magnitude of acceleration vector at each time point, but also the
maximum, the minimum, and the standard deviation of vector
magnitude within a time window. The experiments compared the
performance of four kinds of basic multi-class classifiers and the
performance of four kinds of ensemble learning methods based on
three kinds of basic multi-class classifiers. The results show that
while the method with the highest accuracy is ECOC based on
Random forest.
Abstract: The detection of moving objects from a video image
sequences is very important for object tracking, activity recognition,
and behavior understanding in video surveillance.
The most used approach for moving objects detection / tracking is
background subtraction algorithms. Many approaches have been
suggested for background subtraction. But, these are illumination
change sensitive and the solutions proposed to bypass this problem
are time consuming.
In this paper, we propose a robust yet computationally efficient
background subtraction approach and, mainly, focus on the ability to
detect moving objects on dynamic scenes, for possible applications in
complex and restricted access areas monitoring, where moving and
motionless persons must be reliably detected. It consists of three
main phases, establishing illumination changes invariance,
background/foreground modeling and morphological analysis for
noise removing.
We handle illumination changes using Contrast Limited Histogram
Equalization (CLAHE), which limits the intensity of each pixel to
user determined maximum. Thus, it mitigates the degradation due to
scene illumination changes and improves the visibility of the video
signal. Initially, the background and foreground images are extracted
from the video sequence. Then, the background and foreground
images are separately enhanced by applying CLAHE.
In order to form multi-modal backgrounds we model each channel
of a pixel as a mixture of K Gaussians (K=5) using Gaussian Mixture
Model (GMM). Finally, we post process the resulting binary
foreground mask using morphological erosion and dilation
transformations to remove possible noise.
For experimental test, we used a standard dataset to challenge the
efficiency and accuracy of the proposed method on a diverse set of
dynamic scenes.
Abstract: In this study, data loss tolerance of Support Vector Machines (SVM) based activity recognition model and multi activity classification performance when data are received over a lossy wireless sensor network is examined. Initially, the classification algorithm we use is evaluated in terms of resilience to random data loss with 3D acceleration sensor data for sitting, lying, walking and standing actions. The results show that the proposed classification method can recognize these activities successfully despite high data loss. Secondly, the effect of differentiated quality of service performance on activity recognition success is measured with activity data acquired from a multi hop wireless sensor network, which introduces high data loss. The effect of number of nodes on the reliability and multi activity classification success is demonstrated in simulation environment. To the best of our knowledge, the effect of data loss in a wireless sensor network on activity detection success rate of an SVM based classification algorithm has not been studied before.
Abstract: Sensor-based Activity Recognition systems usually accounts which sensors have been activated to perform an activity. The system then combines the conditional probabilities of those sensors to represent different activities and takes the decision based on that. However, the information about the sensors which are not activated may also be of great help in deciding which activity has been performed. This paper proposes an approach where the sensory data related to both usage and non-usage of objects are utilized to make the classification of activities. Experimental results also show the promising performance of the proposed method.
Abstract: In this paper, the detection and tracking of face, mouth, hands and medication bottles in the context of medication intake monitoring with a camera is presented. This is aimed at recognizing medication intake for elderly in their home setting to avoid an inappropriate use. Background subtraction is used to isolate moving objects, and then, skin and bottle segmentations are done in the RGB normalized color space. We use a minimum displacement distance criterion to track skin color regions and the R/G ratio to detect the mouth. The color-labeled medication bottles are simply tracked based on the color space distance to their mean color vector. For the recognition of medication intake, we propose a three-level hierarchal approach, which uses activity-patterns to recognize the normal medication intake activity. The proposed method was tested with three persons, with different medication intake scenarios, and gave an overall precision of over 98%.
Abstract: Recognizing human action from videos is an active
field of research in computer vision and pattern recognition. Human
activity recognition has many potential applications such as video
surveillance, human machine interaction, sport videos retrieval and
robot navigation. Actually, local descriptors and bag of visuals words
models achieve state-of-the-art performance for human action
recognition. The main challenge in features description is how to
represent efficiently the local motion information. Most of the
previous works focus on the extension of 2D local descriptors on 3D
ones to describe local information around every interest point. In this
paper, we propose a new spatio-temporal descriptor based on a spacetime
description of moving points. Our description is focused on an
Accordion representation of video which is well-suited to recognize
human action from 2D local descriptors without the need to 3D
extensions. We use the bag of words approach to represent videos.
We quantify 2D local descriptor describing both temporal and spatial
features with a good compromise between computational complexity
and action recognition rates. We have reached impressive results on
publicly available action data set
Abstract: This paper addresses the problem of recognizing and
interpreting the behavior of human workers in industrial
environments for the purpose of integrating humans in software
controlled manufacturing environments. In this work we propose a
generic concept in order to derive solutions for task-related manual
production applications. Thus, we are able to use a versatile concept
providing flexible components and being less restricted to a specific
problem or application. We instantiate our concept in a spot welding
scenario in which the behavior of a human worker is interpreted
when performing a welding task with a hand welding gun. We
acquire signals from inertial sensors, video cameras and triggers and
recognize atomic actions by using pose data from a marker based
video tracking system and movement data from inertial sensors.
Recognized atomic actions are analyzed on a higher evaluation level
by a finite state machine.
Abstract: Rapid advancement in computing technology brings
computers and humans to be seamlessly integrated in future. The
emergence of smartphone has driven computing era towards
ubiquitous and pervasive computing. Recognizing human activity has
garnered a lot of interest and has raised significant researches-
concerns in identifying contextual information useful to human
activity recognition. Not only unobtrusive to users in daily life,
smartphone has embedded built-in sensors that capable to sense
contextual information of its users supported with wide range
capability of network connections. In this paper, we will discuss the
classification algorithms used in smartphone-based human activity.
Existing technologies pertaining to smartphone-based researches in
human activity recognition will be highlighted and discussed. Our
paper will also present our findings and opinions to formulate
improvement ideas in current researches- trends. Understanding
research trends will enable researchers to have clearer research
direction and common vision on latest smartphone-based human
activity recognition area.