Publication: Dynamic hand gesture recognition using deep learning for human computer interaction
Loading...
Date
2024-08
Authors
Teng, Peg Gie
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This research develops a dynamic hand gesture recognition system utilizing advanced deep learning techniques to enhance human-computer interaction through
intuitive gesture-based controls. Driven by the need for hands-free interaction with media players, the study's primary objective was to create a real-time system capable of accurately interpreting hand gestures from video sequences. To accomplish this, a sophisticated 3D Convolutional Neural Network (3DCNN) architecture was designed, leveraging both spatial and temporal features extracted from the Jester dataset. This robust 3DCNN model features four convolutional layers followed by fully connected layers, addressing the challenges of variability in hand appearances, illumination changes, and computational complexity. The system's design and implementation were meticulously evaluated through a series of live demonstrations, validating its capability for real-time processing and practical application. The research emphasized a comprehensive analysis of the model's performance metrics, demonstrating its effectiveness in recognizing a wide range of hand gestures and its seamless integration with media playback controls. Key results from the study highlight that the model achieved an impressive accuracy of approximately 93% on the training dataset and 81% on the validation dataset. These results reflect significant advancements in gesture recognition accuracy and real-time processing efficiency, with notable improvements in precision, recall, and F1 scores over multiple epochs. The system’s performance metrics underscore its ability to accurately and efficiently process hand gestures in real-time, enhancing user interaction with media content. The gesture-based control proved to be both engaging and efficient, showcasing the system's practical application potential in enhancing user experience through intuitive, hands-free media player control.