Answers are grounded in this paper and cite their sources.
Shows enhanced face tracking accuracy through CNNs, highlighting technology for disabled individuals.
Introduction: Face identification and tracking have grown in importance in a variety of applications, including security, surveillance, and human-computer interaction. The goal of this paper is to create a real-time face detection and tracking system that uses a 2D convolutional neural network (CNN) model. Methods: The model is trained on a dataset of facial images with accompanying bounding boxes, allowing it to recognize and localize faces in real-time video streams with high accuracy for 3D.VGG16 is a well-known CNN architecture that has shown excellent performance in image classification applications. Additional layers are added to the network to modify VGG16 for face detection: a fully connected layer for classification and two fully connected levels for bounding box regression. Result: Facial expressions are a great way to guide cursor movement, with average accuracies of about 95% and 95.5% for 2D and 3D models respectively. Head movements are an additional intriguing technique with an accuracy rate of about 93% for 2D on average. Eye movements can be used to precisely control the cursor, with average accuracy rates of about 94% and 93.5% for 2D and 3D models respectively. Conclusion: The different CNN models were compared based on the accuracies out of which VGG16 results better. 0.05% and 1.58% accuracy improvement was observed for the 3D model over the 2D model for Facial expressions and head movements respectively.
No takes yet. Share an insight, caveat, or question.
Misra et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: