Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

How to detect any number of lanes? #33

Open
lucasjinreal opened this issue Aug 6, 2020 · 1 comment
Open

How to detect any number of lanes? #33

lucasjinreal opened this issue Aug 6, 2020 · 1 comment

Comments

@lucasjinreal
Copy link

Correct me if am wrong, PINet only trained on a fixed number of lanes isn't it? So that in real scenarios, sometimes we may have more than lanes which model trained on, how to detect all these lanes if actual world have more lanes?

@koyeongmin
Copy link
Owner

When the PINet is trained, the number of lanes is not fixed. You can train the model with the arbitrary number of lanes. However, the provided pre-trained weight was trained on tuSimple dataset, and the maximum number of lanes of the dataset is 5. So I can't be sure that the pre-trained weight will work well on more lanes than 5. Only feature size for embedding is fixed, and it has no major effect on tuSimple dataset. However, if you want to train the model for a lot of lanes, you may need to increase the feature size for more better performance. Thank you!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants