Gesture Recognition in RGB Videos UsingHuman Body Keypoints and Dynamic Time Warping

Gesture recognition opens up new ways for humans to intuitively interact with\nmachines. Especially for service robots, gestures can be a valuable addition to\nthe means of communication to, for example, draw the robot's attention to\nsomeone or something. Extracting a gesture from video data and classifying it\nis a challenging task and a variety of approaches have been proposed throughout\nthe years. This paper presents a method for gesture recognition in RGB videos\nusing OpenPose to extract the pose of a person and Dynamic Time Warping (DTW)\nin conjunction with One-Nearest-Neighbor (1NN) for time-series classification.\nThe main features of this approach are the independence of any specific\nhardware and high flexibility, because new gestures can be added to the\nclassifier by adding only a few examples of it. We utilize the robustness of\nthe Deep Learning-based OpenPose framework while avoiding the data-intensive\ntask of training a neural network ourselves. We demonstrate the classification\nperformance of our method using a public dataset.\n

Paper

Similar papers

© 2026 NYSGPT2525 LLC