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

YOLOV5检测头部 #1

Open
Msunshineboy opened this issue Mar 12, 2023 · 5 comments
Open

YOLOV5检测头部 #1

Msunshineboy opened this issue Mar 12, 2023 · 5 comments

Comments

@Msunshineboy
Copy link

您好,在YOLOV5中FPN-PAN网络输出三个检测头用以检测,fairmot中只有一个,请问这该怎么修改?

@zengjie617789
Copy link
Owner

Select one of the feature map outputs and upsample it to 4.

@Msunshineboy
Copy link
Author

May I ask if it is possible to directly use the 80×80×236 feature map output by PAN for detection and tracking?

@zengjie617789
Copy link
Owner

Given the decode of centernet, the smaller downsample ratio, the more details the network kept.
If you using 8080236(7676255 in yolov5), I‘m afraid it failed to detect.

@Msunshineboy
Copy link
Author

Thank you for your answer. Is there any problem if the layer uses multiple headers? For example, multiple headers of YOLOV5 are used for detection tracing.

@Msunshineboy
Copy link
Author

Hello, if you choose a feature graph to output and upsample it to 4, will pan network be a little redundant?

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