In a technical sense, Android developers frequently use biometric applications, such as face tracking, to support Android mobile devices. To keep the tracking accuracy as high as possible, face tracking utilising a hand-held camera of a mobile device must consider both the camera and the face object’s motion, and must be able to tolerate blurring caused by shaking or considerable movement of the face [5]. Furthermore, mobile devices have numerous constraints in terms of hardware resources, such as computing resources. Because of these constraints, the tracking problem on mobile devices is still a work in progress.

The Android platform can take advantage of the Open Computer Vision library (OpenCV). It’s a programming library that’s mostly focused on real-time computer vision [7]. Viola-Jones detector is provided by OpenCV for detecting several faces in real-time.

Figure: Client-Server Criminal Detection Framework

The face detect-track cycle deals with frames captured in real-time, where N is the number of frames captured. At the first frame and every m frames, the detection kicks in to allow other faces to be recognised and tracked. Face windows are determined in the detection stage, and any faces that have not been found for a long time are removed. The face is kept in every tracking iteration via discrete and efficient features. The facial points to be used in the optical flow function are created using the features extraction method.