Gesture recognition is a part of artificial intelligence in the field of computer vision. With gesture recognition, the computer is able to understand the movements captured on the camera/webcam. The benefits of gesture recognition are many, one of which is what researchers are doing regarding hand-tracking gesture recognition of the human right-hand finger to adjust the volume control on a computer or laptop. Based on this background, this research is intended to apply machine learning developed from the OpenCV and MediaPipe libraries to carry out the process of training and testing finger gestures as gestures to control one of the functions in Windows, one of which is volume control. This process uses the OpenCV Library and MediaPipe because they are capable of multiprocessing with real-time data, so the gesture identification process is faster and more accurate. When the camera/webcam captures the frame of the movement of the human's right-hand finger gesture, an augmentation process is carried out and the provision of keypoint localization landmarks is carried out for each knuckle. In this study, only the fingertip landmarks and index finger landmarks were recognized. Machine learning will perform calculations from the distance between the tip of the thumb and the tip of the forefinger which is used to determine changes in the volume of the sound. From the test results of nine trials with different finger poses, 88.89% was obtained. One of the test results failed to read finger movement gestures, due to the landmark position of the tip of the index finger which was closed with the other fingers.
Copyrights © 2022