Research Article
Heel-Strike and Toe-Off Detection Algorithm Based on Deep Neural Networks Using Shank-Worn Inertial Sensors for Clinical Purpose
Table 2
Parameters of the detection model layers.
| | Layer | Type | Dimensions | Dilation rate | Filters | Kernel size | Stride |
| | Input 1 | Input | | — | — | — | — | | Conv 1 (input 1) | Convolutional | | 1 | 6 | 2 | 1 | | Conv 2 (input 1) | Convolutional | | 2 | 6 | 2 | 1 | | Conv 3 (input 1) | Convolutional | | 4 | 6 | 2 | 1 | | Conv 4 (input 1) | Convolutional | | 8 | 6 | 2 | 1 | | Conv 5 (input 1) | Convolutional | | 16 | 6 | 2 | 1 | | Conv 6 (input 1) | Convolutional | | 1 | 6 | 2 | 1 | | Input 2 | Input | | — | — | — | — | | Conv 1 (input 2) | Convolutional | | 1 | 6 | 2 | 1 | | Conv 2 (input 2) | Convolutional | | 1 | 6 | 2 | 1 | | Dense (concatenate) | Dense | | — | — | — | — | | Output | Softmax | 3 | — | — | — | — |
|
|
