Table 3 Comparison of performance (Accuracy %) between EGAN , other features and feature learning methods
From: Unsupervised feature learning-based encoder and adversarial networks
Dataset | Classifier | Feature extraction | ||||
|---|---|---|---|---|---|---|
RGB | GAN | CNNAE | BiGAN | EGAN | ||
Tea clones | AlexNet | 90.48 | 81.85 | 87.83 | 78.76 | 94.02 |
DenseNet | 80.97 | 75.22 | 80.75 | 82.30 | 94.46 | |
PV corn | AlexNet | 92.46 | 84.12 | 92.86 | 83.73 | 93.25 |
DenseNet | 92.46 | 78.57 | 90.08 | 81.74 | 95.63 | |
PV potato | AlexNet | 89.70 | 83.08 | 91.91 | 85.29 | 97.05 |
DenseNet | 94.85 | 80.14 | 88.97 | 85.29 | 98.52 | |
PV apple | AlexNet | 91.32 | 76.53 | 87.76 | 78.57 | 93.87 |
DenseNet | 89.28 | 69.38 | 87.76 | 77.55 | 95.91 | |
MNIST | AlexNet | 96.10 | 90.96 | 95.30 | 90.20 | 97.70 |
DenseNet | 97.00 | 83.63 | 96.07 | 86.70 | 98.83 | |