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

How to train my own dataset? #55

Closed NARRZ closed 2 years ago

NARRZ commented 2 years ago

Hello, I am very interested in your research results. My own lane line data set has been converted to tusimple style. How can I use my own data set for training? Thank you!

lucastabelini commented 2 years ago

Thank you for your interest. You might also be interested in my latest lane detection work: https://github.com/lucastabelini/LaneATT

If your dataset is in TuSimple's format, you can train a model on it using the same procedure described on the README.md: https://github.com/lucastabelini/PolyLaneNet#training

One of the settings that you will have to change is the root of the dataset (which points to "datasets/root" by default). You will also want to modify the number of epochs and the learning rate scheduler.

NARRZ commented 2 years ago

Thank you for your reply. I found that I did not really convert my dataset to tusimple format, and I did not find the correct conversion method, do you have any suggestions to convert the data set to tusimple format? Thank you!

lucastabelini commented 2 years ago

See #5.