Table 3 Comparison of model deployment feasibility.
From: Scene image visual layout based on deep encoder–decoder network and visual image attention model
Task number | Energy costs (Wh) | Memory footprint (GB) | ||||
|---|---|---|---|---|---|---|
DEDN-VGAT LayoutGen | TS-SLG | SA-GAT | DEDN-VGAT LayoutGen | TS-SLG | SA-GAT | |
3 | 61.91 | 82.17 | 68.47 | 6.69 | 7.58 | 7.45 |
6 | 45.77 | 73.02 | 67.18 | 5.92 | 6.59 | 7.36 |
9 | 48.44 | 56.66 | 72.85 | 4.86 | 5.60 | 6.12 |
12 | 61.14 | 83.49 | 71.87 | 5.47 | 6.59 | 7.41 |
15 | 64.11 | 81.47 | 64.95 | 5.05 | 6.15 | 6.59 |
18 | 57.67 | 60.81 | 66.39 | 6.73 | 6.10 | 6.76 |
21 | 45.18 | 73.44 | 74.23 | 5.65 | 7.85 | 6.11 |
24 | 51.04 | 65.69 | 62.93 | 6.80 | 5.67 | 6.97 |
27 | 56.62 | 66.56 | 70.77 | 6.47 | 5.60 | 6.57 |
30 | 57.32 | 84.11 | 68.39 | 5.91 | 8.21 | 6.71 |
Means | 54.92 | 72.74 | 68.80 | 5.96 | 6.59 | 6.81 |
