TABLE 2.
Structure of CNN model with ACC + GYR as input.
| Layer (type) | Size | Output shape | Param # |
|---|---|---|---|
| conv2d (Conv2D) | 18*1 | (None, 100, 6, 32) | 608 |
| max_pooling2d (MaxPooling2D) | 2*1 | (None, 50, 6, 32) | 0 |
| conv2d_1 (Conv2D) | 9*1 | (None, 50, 6, 64) | 18,496 |
| conv2d_2 (Conv2D) | 3*1 | (None, 50, 6, 128) | 24,704 |
| max_pooling2d_1 (MaxPooling2D) | 2*1 | (None, 25, 6, 128) | 0 |
| conv2d_3 (Conv2D) | 1*6 | (None, 25, 1, 128) | 98,432 |
| flatten (Flatten) | -- | (None, 3,200) | 0 |
| dense (Dense) | -- | (None, 3) | 9,603 |
Total params: 151,843; Trainable params: 151,843; Non-trainable params: 0.