Table 3 Computational and parametric quantities comparison of experiment results on CDD and WHU-CD.
From: DASUNet: a deeply supervised change detection network integrating full-scale features
Method type | Network | Params(M) | Gflops(G) | CDD | WHU-CD | ||
|---|---|---|---|---|---|---|---|
Train_Epoch(S) | Test_Epoch(S) | Train_Epoch(S) | Test_Epoch(S) | ||||
CNN | FC-EF | 1.35 | 3.58 | 578.95 | 61.62 | 354.50 | 18.25 |
FC-Siam-conc | 1.55 | 5.33 | 676.39 | 65.77 | 351.09 | 17.52 | |
FC-Siam-diff | 1.35 | 4.73 | 685.02 | 70.45 | 356.32 | 17.25 | |
L-UNet | 8.45 | 17.33 | 870.42 | 98.04 | 443.59 | 23.78 | |
IFNet | 35.99 | 82.27 | 893.78 | 115.45 | 450.91 | 28.96 | |
SNUNet-24 | 6.77 | 30.90 | 873.24 | 103.89 | 470.66 | 26.31 | |
USSFC-Net | 1.52 | 4.86 | 830.91 | 90.23 | 430.95 | 23.44 | |
TinyCD | 0.29 | 1.54 | 704.65 | 80.57 | 329.30 | 20.92 | |
DASUNet-32 | 2.27 | 25.61 | 1022.98 | 135.88 | 515.98 | 34.12 | |
DASUNet-64 | 9.07 | 100.93 | 1731.19 | 226.62 | 874.34 | 56.94 | |
Transformer | ChangeFormer | 29.84 | 11.65 | 688.44 | 77.17 | 347.97 | 17.81 |
IDET | 45.09 | 124.19 | 1232.82 | 126.14 | 669.25 | 32.28 | |
ScratchFormer | 36.92 | 196.59 | 2690.59 | 291.41 | 1387.35 | 71.66 | |
Swin-UNet-CD | 27.15 | 7.75 | 534.23 | 70.77 | 316.49 | 18.63 | |
