Research Article
Intelligent Image Recognition System for Marine Fouling Using Softmax Transfer Learning and Deep Convolutional Neural Networks
Table 6
CNN models training results.
| Data | Number of images | Training images | Validating images | Model name | Epochs | Batch size | Loss | Accuracy | Val_ loss | Val_acc |
| Dataset 1 | 1423 | 997 | 426 | CNN_Model.h1 | 100 | 96 | 0.0166 | 0.9354 | 0.0803 | 0.6879 | CNN_Model.h2 | 75 | 0.0278 | 0.8864 | 0.0743 | 0.6848 | CNN_Model.h3 | 50 | 0.0475 | 0.8062 | 0.0902 | 0.6303 | CNN_Model.h4 | 25 | 0.0669 | 0.7206 | 0.0673 | 0.7091 |
| Dataset 2 | 582 | 402 | 180 | CNN_Model.h5 | 100 | 96 | 0.0191 | 0.9225 | 0.1046 | 0.6250 | CNN_Model.h6 | 75 | 0.0242 | 0.9182 | 0.1041 | 0.6250 | CNN_Model.h7 | 50 | 0.0302 | 0.8916 | 0.0670 | 0.6771 | CNN_Model.h8 | 25 | 0.0617 | 0.7351 | 0.0836 | 0.6146 |
|
|