A sliding window approach to natural hand gesture recognition using a custom data glove

Investigating Interactions for Text Recognition using a Vibrotactile Wearable Display

“…The mapping is natural, as it corresponds to movement of focus point within the sentence. Additionally, the interactions are easy to perform and considered in the research community as natural hand gestures [42,15,26] meaning that most users would be familiar with them. As for current word interaction, we map it to swipe up gesture as it shares the simplicity and popularity with the other selected gestures.…”

“…Gesture recognition has been a subject of many researchers. At a core abstract level, the main approaches for gesture recognition has been based on either environmental sensors such as: camera [8,6,12,8,1,18], radar signals [10,25] wifi [23,36], etc... or hand/body-worn sensors such as : motion sensors, flexion and pressure sensors [26,32,56,33,57]. Even though each of them has its advantages and disadvantages, for our specific use case, wearable based sensors approach is a more suitable solution.…”

“…Many gesture recognition systems using wearable sensor have been proposed over the years, most of them rely on either hand worn sensors [26,32,56,33,57] or wrist-based sensors [17,50,55,58]. For recognizing gestures, typically statistical and machine learning approaches such as neural networks [32,56], support vector machines [26], linear discriminant analysis (LDA) [26,15], logistic regression [26,58], decision trees [58], etc.. have been employed.…”

“…Our gesture recognition system uses an existing framework and dataset of Luzhnica et al [26]. The authors, collected data from 18 participants performing 31 gestures using a custom made data glove, where each participant performed each gesture 5-10 times.…”

Exploring in-the-Wild Game-Based Gesture Data Collection

“…For example, gesturebased interaction can benefit interaction in areas where touch input is not suitable. However, to develop gesture-based interfaces, UI designers may incorporate gesture recognition systems, often requiring a large corpus of training data [1][2][3]. In many cases, it is difficult to efficiently collect a large corpus of training data in a short period of time, representing a limitation in the use of gesture elicitation controllers.…”