PyTorch implementation of "GhostNet: More Features from Cheap Operations"
Concept ot the GhostNet [1].
| Indicator | Value |
|---|---|
| Accuracy | 0.99230 |
| Precision | 0.99233 |
| Recall | 0.99221 |
| F1-Score | 0.99225 |
Confusion Matrix
[[ 976 0 0 0 0 1 3 0 0 0]
[ 0 1135 0 0 0 0 0 0 0 0]
[ 2 1 1027 0 0 0 0 2 0 0]
[ 0 1 0 998 0 10 0 0 1 0]
[ 0 0 1 0 977 0 0 1 0 3]
[ 1 0 0 3 0 887 1 0 0 0]
[ 3 6 1 0 0 1 947 0 0 0]
[ 2 8 0 1 2 0 0 1015 0 0]
[ 2 1 2 1 1 5 0 0 961 1]
[ 3 1 0 1 4 0 0 0 0 1000]]
Class-0 | Precision: 0.98686, Recall: 0.99592, F1-Score: 0.99137
Class-1 | Precision: 0.98439, Recall: 1.00000, F1-Score: 0.99213
Class-2 | Precision: 0.99612, Recall: 0.99516, F1-Score: 0.99564
Class-3 | Precision: 0.99402, Recall: 0.98812, F1-Score: 0.99106
Class-4 | Precision: 0.99289, Recall: 0.99491, F1-Score: 0.99390
Class-5 | Precision: 0.98119, Recall: 0.99439, F1-Score: 0.98775
Class-6 | Precision: 0.99579, Recall: 0.98852, F1-Score: 0.99214
Class-7 | Precision: 0.99705, Recall: 0.98735, F1-Score: 0.99218
Class-8 | Precision: 0.99896, Recall: 0.98665, F1-Score: 0.99277
Class-9 | Precision: 0.99602, Recall: 0.99108, F1-Score: 0.99354
Total | Accuracy: 0.99230, Precision: 0.99233, Recall: 0.99221, F1-Score: 0.99225
- PyTorch 1.11.0
[1] Han, Kai, et al. "Ghostnet: More features from cheap operations." Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020.