Real-time human face tracking is suggested in this paper for use while interacting with robots. The system consists of two main parts, the first works to discover the human face, determine its location in relation to the original image, and find the dimensions of the face to be used in the second part of this system. The second part receives the location and dimension of the face and tracks it by controlling the movement of the camera according to the offset between the interval. In order to detect human faces for the earlier job, the Haar cascade method is used, whereas the Kanade-Lucas-Tomasi (KLT) algorithm is used for face tracking under various circumstances. As a trace output from the prior stage, the camera is offset by its offset between image frames. The findings of the experiments demonstrate that real-time tracking of human faces was successful even when the participants were donning glasses, hats, or face-side positions. At a maximum frame rate of 26 fps, experiments were conducted.