Analysis of Touchless Mouse Technology for Physical Disabilities

Abstract

We provide a touchless mouse system that surpasses past attempts by utilising deep learning models such as DenseNet169 and DenseNet201, in addition to an ensemble model. For feature extraction in the touchless mouse system, we use two different state-of-the-art convolutional neural network architectures, namely DenseNet169 and DenseNet201. These models, which were trained using massive datasets, perform remarkably well regarding computer vision tasks. Touchless mouse technology's sophisticated feature extraction capabilities make the exact recognition and interpretation of hand motions and movements possible. An ensemble model is developed by integrating the results of DenseNet169 and DenseNet201. This is done to make the system's performance even more effective. The ensemble technique improves the accuracy, stability, and generalizability of hand gesture detection by capitalising on these distinctions and using them to its advantage. Comparisons are made between the DenseNet169 and DenseNet201 models, the Ensemble model and several other deep learning and ensemble learning models. Additional deep learning and ensemble learning models are also displayed. The Ensemble model reached the maximum attainable accuracy of 99.62 per cent.