Research Article
Infrared Thermal Image Gender Classifier Based on the Deep ResNet Model
Table 3
The Performance measures of the proposed IRT_ResNet models.
| Net name | Metrics | Data 1 ratio (%) | Data 2 ratio (%) | All data ratio (%) |
| IRT_ResNet 18 | Precision | 96 | 92 | 96.6 | Recall | 94 | 99 | 93.8 | F-score | 95.3 | 95.7 | 95.2 | Overall accuracy | 95.4 | 95.6 | 95.2 | IRT_ResNet 50 | Precision | 96 | 97 | 94.5 | Recall | 87 | 98.6 | 96 | F-score | 91.5 | 97.8 | 95.3 | Overall accuracy | 91.9 | 97.8 | 95.2 | IRT_ResNet 101 | Precision | 85 | 99 | 97.3 | Recall | 97.7 | 94.3 | 97 | F-score | 90.9 | 96.6 | 97.2 | Overall accuracy | 90.2 | 96.7 | 97.2 |
|
|