Hand Sign Classification Techniques Using Supervised And Unsupervised Learning Algorithms

Gesture recognition is becoming a sore subject in computer vision captivating the idea of social interaction not only between humans but also create ergonomic systems that control devices ranging from time-of-flight cameras and controlling vehicles to virtual reality. Even though it is gaining ground recently, fast and robust recognition remains an unsolved problem. To our understanding, we are utilizing the Leap motion sensor captured near-infrared image dataset for gesture identification, which tides over word-level sign recognition encompassing a diverse set of hand gestures. We have suggested the approach of extracting the gesture from the image using PCA and image segmentation followed by the feature extraction stage. In this paper, we have analyzed and differentiated various methods of hand recognition counting Random forests, stochastic gradient descent, Naive Bayes, Decision tree, and Logistic regression algorithms and exploited our results