This work presents the development of a computer vision solution for the quantitative assessment of facial movements, with the aim of supporting the diagnosis and rehabilitation of facial paralysis. The system, implemented in Python using the OpenCV and MediaPipe libraries, detects 468 facial landmarks and performs geometric calculations to monitor the amplitude and symmetry of expressions in real-time, using only a conventional webcam without depth sensors. Preliminary results demonstrate the tool's capability to translate vector coordinates into clinical metrics and instant visual feedback, establishing a foundation for future gamification applications in virtual environments aimed at stimulating neuroplasticity and patient engagement in physical therapy treatment.