Research Article
Driver Distraction Identification with an Ensemble of Convolutional Neural Networks
Table 3
Distraction classification results by training (AlexNet) and testing on skin segmented images.
| Network | Accuracy [%] |
| No pixels’ spatial info | 93.14 |
| Spatial info learned through active learning with 50% of data | 93.53 |
| Spatial info learned through active learning with 100% of data | 93.62 |
|
|