search

lucastabelini / PolyLaneNet

Code for the paper entitled "PolyLaneNet: Lane Estimation via Deep Polynomial Regression" (ICPR 2020)
https://arxiv.org/abs/2004.10924
MIT License
301 stars 75 forks source link

Check param count #54

Closed voldemortX closed 3 years ago

voldemortX commented 3 years ago

@lucastabelini Hi again! I ran a parameter count with thop on PolyLaneNet (EfficientNetB0), and got 0.87M, do you have a official number recorded on this? I'm not sure if thop is correct about EfficientNet, since the mac count is wrong (< 1G).

I copied speed.py from LaneATT.

lucastabelini commented 3 years ago

It seems thop doesn't give accurate results for EfficientNet (you can see on its output that it ignores many kinds of layers). However, there's a simpler way to get the parameter count:

pytorch_total_params = sum(p.numel() for p in model.parameters())

This gives me around 4M parameters.

voldemortX commented 3 years ago

Thanks! I got it.