Energy Efficient Activity Recognition Based on Low Resolution Accelerometer in Smart Phones

Liang, Yunji; Zhou, Xingshe; Yu, Zhiwen; Guo, Bin; Yang, Yue

doi:10.1007/978-3-642-30767-6_11

Cited by 31 publications

(24 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…And there is no optimal solution still existed for an adaptive HAR system. Several studies [12][13][14][15][16] have attempted to tackle various issues in this field, for example, position-or orientation-independent activity recognition.…”

Section: Related Workmentioning

confidence: 99%

Adaptive HAR System to Improve Recognition Accuracy

Awan

2018

Mehran Univ. res. j. eng. technol.

View full text Add to dashboard Cite

HAR (Human Activity Recognition) system becomes complex, inefficient and less accurate as we keep on adding new activities into the system; because it follows a specific procedure for activity recognition, from raw data collection to classification. In this study, we discuss an adaptive system to improve recognition accuracy. We developed a mathematical model to categorize the activities based on their data pattern. It observed that as we group the activities; although a separate classification model is required for each group, but it increases the recognition accuracy and efficiency of the system. The experiments on the data of eleven activities gathered from 10 volunteers proved the usability, scalability and effectiveness of our proposed methodology. The recognition accuracy of eleven activities was increased in total about 9-37% and reached up to 90% in different cases, using different number of groups and classification algorithms.

show abstract

Section: Related Workmentioning

confidence: 99%

Adaptive HAR System to Improve Recognition Accuracy

Awan

2018

Mehran Univ. res. j. eng. technol.

View full text Add to dashboard Cite

show abstract

“…However, these should not compromise the recognition performance. In order to see how these two parameters affect the CPU usage, we evaluated our app with three different window sizes (10, 5, 2 seconds) and three sampling rates ( 50,20,10). We measure the CPU usage of our app on the phone as well as we measure CPU time of the activity recognition pipeline (AR Pipeline with SMO).…”

Section: Impact Of Sampling Rate and Window Sizementioning

confidence: 99%

“…Most of the work in this area is performed offline such that collected data is analyzed in machine learning tools such as WEKA, Scikit-learn, R, and MATLAB. Recently, researchers have been moving towards online activity recognition in order to verify the offline results and to analyze the resource consumption of machine learning algorithms on mobile phones and other wearable devices, such as smartwatches [1], [3], [5], [6], [10]- [12].…”

Section: Introductionmentioning

confidence: 99%

Resource consumption analysis of online activity recognition on mobile phones and smartwatches

Shoaib

İncel

Scolten

et al. 2017

2017 IEEE 36th International Performance Computing and Communications Conference (IPCCC)

View full text Add to dashboard Cite

Abstract-Most of the studies on human activity recognition using smartphones and smartwatches are performed in an offline manner. In such studies, collected data is analyzed in machine learning tools with less focus on the resource consumption of these devices for running an activity recognition system. In this paper, we analyze the resource consumption of human activity recognition on both smartphones and smartwatches, considering six different classifiers, three different sensors, different sampling rates and window sizes. We study the CPU, memory and battery usage with different parameters, where the smartphone is used to recognize seven physical activities and the smartwatch is used to recognize smoking activity. As a result of this analysis, we report that classification function takes a very small amount of CPU time out of total app's CPU time while sensing and feature calculation consume most of it. When an additional sensor is used besides an accelerometer, such as gyroscope, CPU usage increases significantly. Analysis results also show that increasing the window size reduces the resource consumption more than reducing the sampling rate. As a final remark, we observe that a more complex model using only the accelerometer is a better option than using a simple model with both accelerometer and gyroscope when resource usage is to be reduced.

show abstract

“…The testing scenarios for all of these studies are summarized in Table 2.10, where NA stands for not available. For some studies, there was no information available on how they tested their implemented systems [51,78,69,63,73,53,54,71]. …”

Section: Validation Of Online Activity Recognitionmentioning

confidence: 99%

“…Though many studies simply report the number [63] A1, A4, A5, A12 similarity score using geometric template matching algorithm Android phones [11] A1, A5, A8, A10 mean, VAR Android Nexus One [72] A1, A2, A3, A5, A6 mean, root mean square, difference between max and min values Android phones [73] Different physical activities maximum and minimum euclidean norm ZTE Blade [77] A1, A2, A3, A4, A5, A9 signal magnitude, coefficient of variance, counts per minute Samsung Galaxy S [68] A1, A2, A3, A5 mean, min, max, SD Samsung Galaxy Gio [50] Decision tree + DHMM Symbian Nokia N95 A, GPS 4.72 Lu et al [54] Decision tree Symbian, iOS Nokia N95, iPhone A iPhone (0.9-3.7), N95 (1-3) Berchtold et al [78] Fuzzy classification Debian Linux OpenMoko Neo Freerunner A 3.3 Lane et al [11] Naive Bayes Android Android Nexus One A 11 Siirtola [70] QDA Symbian, Android Samsung Galaxy Mini, Nokia N8 A 5 Kose et al [68] Naive Bayes, KNN clustered Android Samsung Galaxy Gio A 42 (Naive), 29 (KNN) Siirtola and Roning. [57] Decision tree Symbian Nokia N8 A 15 of hours a battery lasts as a resource metric [77,52,53,11,54,70,60], it has a drawback. Many of these studies use different mobile phones with different battery capacities, so this metric can be misleading.…”

Section: Resource Consumption Analysismentioning

confidence: 99%

Sitting is the new smoking: online complex human activity recognition with smartphones and wearables

Shoaib¹

View full text Add to dashboard Cite

Energy Efficient Activity Recognition Based on Low Resolution Accelerometer in Smart Phones

Cited by 31 publications

References 14 publications

Adaptive HAR System to Improve Recognition Accuracy

Adaptive HAR System to Improve Recognition Accuracy

Resource consumption analysis of online activity recognition on mobile phones and smartwatches

Sitting is the new smoking: online complex human activity recognition with smartphones and wearables

Contact Info

Product

Resources

About