Table 2 Summary of the presented classification models.
From: Collection of 2429 constrained headshots of 277 volunteers for deep learning
Label | Method | Number of headshots | Pixels per sample | Augmentation | Cross-vallidation | Accuracy | F-score |
|---|---|---|---|---|---|---|---|
Female | Jubatus | 277 | 75 | 1× | Leave-one-voluneteer-out | 0.765 | 0.767 |
Female | Random forest | 277 | 75 | 1× | Leave-one-voluneteer-out | 0.791 | 0.803 |
Female | LightGBM | 277 | 75 | 1× | Leave-one-voluneteer-out | 0.801 | 0.812 |
Female | Random forest | 2429 | 299 × 299 | 10× | Leave-one-voluneteer-out | 0.8600 | 0.8671 |
Female | CNN | 2429 | 299 × 299 | 10× | Leave-one-voluneteer-out | 0.9642 | 0.9621 |
Smilling | CNN | 668 × 2 | 299 × 299 | 10× | 10-fold | 0.9303 | 0.9297 |
