Convolutional Neural Network Based Virtual Mouse

Individuals with disabilities frequently encounter difficulties when utilizing computers due to the prevalent mode of interaction relying on a mouse and keyboard. This conventional method can pose significant challenges for individuals with physical impairments. However, facial movements offer a promising alternative, as they can be used to operate a computer through facial expression recognition. This approach, known as face recognition technology, represents a contemporary form of human-computer interaction (HCI) that could potentially replace conventional HCI technologies in the future.This research study focuses on designing, implementing, and evaluating experiments related to facial gesture recognition for computer cursor control. Specifically, we train a ResNet18 convolutional neural network model using a dataset for landmark detection to accurately recognize facial movements. The results obtained are presented, and the potential benefits and ethical considerations of this technology are discussed. The ultimate goal of this research is to improve the lives of people with disabilities by providing them with a more accessible and efficient means of interacting with computers.