Table 9 Comparison of time complexity and parameter sizes for different modules.
From: A simple monocular depth estimation network for balancing complexity and accuracy
Module | Time complexity | Params |
|---|---|---|
Standard Transformer13 | \(O(H^2 \times W^2 \times C_{in})\) | 4.09M |
DCF (Ours) | \(O(H \times W \times C_{in}^2)\) | 1.54M |
Ordinary Convolution | \(O(H \times W \times K^2 \times C_{in} \times C_{out})\) | 65.79K |
DyConv41 | \(O(H \times W \times K^2 \times C_{in} \times C_{out} \times N)\) | 264.22K |
LMC (Ours) | \(O(H \times W \times K^2 \times C_{in} \times C_{out} \times N)\) | 287.84K |
MHSA13 | \(O(H^2 \times W^2 \times C_{in})\) | 263.17K |
W-MSA35 | \(O(H \times W \times C_{in}^2 + N_w \times W_s^4 \times C_{in})\) | 66.05K |
WAT (Ours) | \(O(H^2 \times W^2 \times C_{in} / 16)\) | 887.23K |
