MyProject: Deep Neural Networks with Physical Layers
Jonathan Perera Jake Cariello
jjp56@duke.edu jec91@duke.edu
| Paper PDF |
|
The need to identify and categorize hand gestures is becoming more prevalent as many facets of society become more and more digitalized. Using mean optical flow data of 6 distinct hand gestures, we use various preprocessing methods followed by a novel convolutional neural network to efficiently categorize these hand gestures with high accuracy. In addition, we visualize both our input data set and results to draw inferences about what the most important parts of the data set are with respect to distinguishing between gestures. |
|
|
| Paper: |
Code and Data:
|