Research Article

Towards a Complete Set of Gym Exercises Detection Using Smartphone Sensors

Table 8

Comparison table for KNN classification results at three body positions (exercisewise) of the only accelerometer and both accelerometer and gyroscope sensors.

Exercise groupExercise nameThree accelerometers’ accuracy (%)Three accelerometers + two gyroscopes’ accuracy (%)Difference (%)Avg. time (in sec) consumed at exercise, 2 sets each of 10 reps

Unique exercises groupFace pull1001000.026.8
Cable front raise1001000.038.5
Scott curl98.499.2+0.633.2
Smith machine drag curl95.6100+4.430.9
Triceps with bar98.199.0+0.926.8
Decline close grip bench press1001000.026.3
Standing cable cross1001000.028.7
Wide grip pull up99.599.7+0.223.8
T bar rows1001000.024.2
Chin ups1001000.023.7
Adjustable sit-up bench1001000.037.9
Abs wheel1001000.046.5
Roman chairs1001000.037.8
Flutter kick1001000.028.7
Common exercises groupSeated barbell shoulder press10097.4−2.627.1
Incline press wide grip99.1100+0.926.6
Standing barbell press97.099.1+2.927.4
Inclined dumbbell curl1001000.037.2
Barbell preacher curl99.399.30.034.4
Triceps press with cable98.999.0+0.129.2
Machine bench press1001000.025.2
Leg press1001000.041.3
Romanian deadlift1001000.037.2
Barbell squat1001000.030.2
Dips1001000.025.3
Butterfly99.5100+0.541.0
Both groupsAll exercises average accuracy99.5199.72+0.2131.4