Piezoelectric force touch panels are extensively utilized as human-machine interfaces for 3-dimensional touch sensing in internet of things (IoT) applications. However, the unstable force voltage responsivity issue induced by different touch orientations limits the successful use of piezoelectric touch panels. In this article, a piezoelectric touch panel, which is sensitive to both capacitive and force stimulation, is assembled; and a touch orientation classification technique is developed to calibrate the detected force amplitude by training a machine learning model with finger induced capacitive information. Finally, a high stable force voltage responsivity of 87.5% is achieved experimentally.