Gesture classification with machine learning using Kinect sensor data

Human Activity Recognition for Physical Rehabilitation

“…In addition, results presented in this paper illustrate that RF offers performance advantages compared to SVM in the classification of activities based on our 3D kinematic dataset. As in [10], we also note that there has to be an understanding of the anatomical differences between training and testing participants which could influence the result if no prior normalisation is undertaken.…”

“…Bhattacharya et al [10] analysed SVM and Decisions Trees (DT) for how they perform in detecting and classify gestures used in aircraft marshalling. The study found that SVM outperformed DT, though the authors noted that DT were susceptible to participant anatomical differences whereas SVM 978-1-4799-0652-9/13/$31.00 c 2013 IEEE was participant independent and not affected by differences in body posture.…”

Motion Prediction in Rock Scissor Paper Game based on Machine Learning

“…For realizing this, the research should select specific data length within the whole of data length which would be used for discerning gesture data and implement converting data formation for machine learning. [5] Furthermore, a large amount of data as to rockpaper-scissors gesture are required to the more accurate recognition of user gesture. Data will be processed as the formation in below figure6.…”

A Simple Manner of Dynamic Gesture Recognition Based on Kinect

“…In 2010, Microsoft's Kinect Sensor was published, and the characteristics of its low cost and high efficiency make it preferred in gesture recognition technology research. For example, Raheja, Chaudhary and Singal use the Kinect sensor to realize the track of the fingers and the palm center; Bhattacharya [4], Bhattacharya and Czejdo utilize support vector machine (SVM) and the decision tree algorithm for finishing gesture recognition in aviation flight [5]; Luo, Xie and Zhang apply Kinect sensor for depth information combined with hidden Markov model (HMM) to identify five dynamic gesture trajectories to control the movement of wheelchair [6]. This paper firstly puts forward the concept of "micro-gestures" to break down of dynamic hand gesture trajectories which is from Kinect for depth information and then combines with dynamic time warping (DTW) algorithm to recognize dynamic gestures.…”