Dynamic hand gestures have become increasingly popular as an input modality for interactive systems. There exists a variety of arm-worn devices for the recognition of hand gestures, which differ not only in their capabilities, but also in the arm positions where they are worn. The aim of this paper is to investigate the effect of placement of such devices on the accuracy for recognizing dynamic hand gestures (e.g. waving the hand). This is relevant as different devices re- quire different positions and thus differ in the achievable recognition accuracy. We have chosen two positions on the forearm: on the wrist and right below the elbow. These po- sitions are interesing as smartwatches are usually worn on the wrist and devices using EMG sensors for the detection of static hand gestures (e.g. spreading the fingers) have to be worn right below the elbow. We used an LG G Watch worn on the wrist and a Myo armband from Thalmic Labs worn below the elbow. Both are equipped with three-axis accelerometers, which we used for gesture recognition. Our hypothesis was that the wrist- worn device would have a better recognition accuracy, as dynamic hand gestures have a bigger action radius on the wrist and therefore lead to bigger acceleration values. We conducted a comparative study with nine participants that performed eight simple, dynamic gestures on both devices. We tested the 4320 gesture samples with different classifiers and feature sets. Although the recognition results for the wrist-worn device were higher, the difference was not signif- icant due to the substantial variation across participants.

  author = {Kefer, Kathrin and Holzmann, Clemens and Findling, Rainhard Dieter},
  title = {Comparing the Placement of Two Arm-Worn Devices for
  		  Recognizing Dynamic Hand Gestures},
  booktitle = {Proc. {MoMM} 2016: 14th International Conference on
  		  Advances in Mobile Computing and Multimedia},
  year = {2016},
  pages = {99--104},
  address = {Singapore},
  month = nov,
  publisher = {ACM},
  doi = {10.1145/3007120.3007146},
  keywords = {Gesture recognition; Hand gestures; Accelerometer; Sensor
  		  placement; Smartwatch;Myo armband}