The facial emotion recognition ecosystem utilises computer vision and deep learning techniques and has emerged as a crucial study area in mental health applications. This research investigates enhancing facial expression recognition performance through fine-tuning of Vision Transformer (ViT) models, known for their adaptability in capturing global context and long-range image dependencies. The research methodology involves fine-tuning the pre-trained Google ViT model on the FER2013 dataset of grayscale facial expression images categorised into seven emotions. Data augmentation techniques are employed with stratified datasets to improve model generalization. The increasing demand for an information system (IS) guided vision and automated motion analysis has motivated us to develop fine-tuning models that meet user needs and recommendations. The purpose of the study is to create fine-tuning vision transform models. The fine-tuned ViT model, with its adaptability and effectiveness, achieves an accuracy of 77.09% in recognising facial expressions, demonstrating its potential in visual applications.