This paper addresses the limited digital accessibility faced by individuals with reduced hand mobility and explores how to overcome it. An efficient facial gesture recognition model is proposed, designed to operate based on facial landmarks (Facial Mesh) extracted using Mediapipe’s Face Mesh technology. The model incorporates a graph neural network (GNN) with a lightweight architecture and low computational complexity to detect facial gestures on the user’s face, enabling smooth execution on low-performance computers without significantly compromising user experience. The work presented may be of interest to researchers focused on digital accessibility for individuals with motor impairments in their hands.