Fd-MobileNet

Links

  1. Paper
  2. Code Implementation
  3. MobileNet and FD-MobileNet models

Notes

comparison

  • FD (Fast Downsampling) downsamples early. This means there is less operations early on, more operations after downsampled when feature map is smaller.

FD-MobileNet x0.25 only has 0.383M params at 43.81% top-1 accuracy compared to MobileNet x0.25 with 0.47M params at 54.22% accuracy. MobileNet seems way more accurate, but we are really hardware limited, so I think this is promising.

code

  • Only changes are number of channels and stride sizes!