Table 4.
SN | Attributes | PSP-Net | VGG-SegNet | VGG-UNet | ResNet-SegNet | ResNet-UNet |
---|---|---|---|---|---|---|
1 | Backbone-encoder | NA | VGG-16 | VGG-16 | Res-50 | Res-50 |
2 | # Parameters | ~4.4 M | ~11.6 M | ~12.4 M | ~15 M | ~16.5 M |
3 | # NN layers | 54 | 33 | 36 | 160 | 165 |
4 | Model size (MB) | 50 | 133 | 142 | 171 | 188 |
5 | Batch size | 8 | 8 | 4 | 4 | 4 |
6 | Training time * | ~15 | ~50 | ~54 | ~60 | ~63 |
7 | Prediction time | <1 s | <1 s | <1 s | <1 s | <1 s |
* In minutes; MB: megabytes; M: million; NN: neural network; Res: ResNet.