Files in this item

FilesDescriptionFormat

application/pdf

application/pdfECE499-Sp2015-aranguiz.pdf (441kB)
(no description provided)PDF

Description

Title:RecoLift: An Android Wear Fitness Tracker for Strength Training
Author(s):Aranguiz, Dario
Contributor(s):Choudhury, Romit R.
Subject(s):Android
strength training
wearable
inertial sensors
Abstract:Despite the plethora of fitness trackers on the market, few monitor signals other than number of steps and heart rate. With the increasing mainstream acceptance of general-purpose smartwatches, however, we have the capability to track more complex activities. We propose RecoLift, an Android-based system to track exercises and repetitions in weight training and bodyweight training activities based on the work of Morris et al. (2014). Our goal is a system which provides feedback to the user in an autonomous, online fashion, harnessing both smartwatch and smartphone sensors. This system is separated into three key phases: segmentation, during which we use the periodicity of the signals to determine if an exercise is being performed, recognition, which calculates signal features to determine which exercise is being performed, and counting, which uses periodicity to calculate the number of repetitions in a set. Our evaluations trained on a single user show perfect exercise recognition, with 80% of repetitions on average being within 1 repetition of the true count.
Issue Date:2015-05
Genre:Other
Type:Text
Language:English
URI:http://hdl.handle.net/2142/78982
Date Available in IDEALS:2015-08-03


This item appears in the following Collection(s)

Item Statistics