Gesture-based Television Controller Using PIR Sensor Array
Manuel Luis C. Landrito, Axcel Justin E. Abanilla, Noel B. Linsangan · 2024
Elderly people and those with disabilities may have difficulty with physical activities such as using electronics. For the study, the researchers will identify the gestures needed, use CNN to convert hand gestures to TV commands and test its effectiveness. A grid of 4x4 PIR sensors is arranged to help identify hand swipes. These hand swipes are then used to control the television. The television will turn on or off, switch channels, mute, and change the volume. Data was gathered by doing 50 hand swipes each in eight directions, namely up, left, right, down, up right, up left, down right, and down left. This data was inputted into a Keras 1-Dimensional Convolution Neural Network (1D CNN) for training. The resulting general accuracy is 55%. However, the device can recognize hand swipes and turn them into TV commands. To further enhance this, the researchers recommend reducing the number of directions to up, down, left, and right and comparing it with other machine learning algorithms.