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
295 stars 74 forks source link

h_samples #57

Closed liangjxiong closed 2 years ago

liangjxiong commented 2 years ago

Hello! The number of h_samples in my own dataset is 72. Can I run it with your code? If it can be executed, how should I fix your modified code? thank you!

lucastabelini commented 2 years ago

If your dataset's annotations are in TuSimple's format, it should not be difficult. You can modify the default settings to use your data.

liangjxiong commented 2 years ago

Can you specify which parameters should be modified? I don't know much about this.

liangjxiong commented 2 years ago

Another point is that the maximum number of lane lines in my dataset is 4, and some are less than 4. How should this be solved?

liangjxiong commented 2 years ago

I have solved the above problems. I run: Python test py --exp name tusimple resnet34 --cfg ./ cfgs/tusimple_ resnet34. yaml --epoch 2695。FPS is 947.644. Why is FPS so high? In your paper, FPS is more than 100.

lucastabelini commented 2 years ago

I'm not sure. But if the output seems correct, the faster the better :)