Skip to content

Conversation

@tianzhi0549
Copy link
Member

The PR fixes a demo bug of FCOS. @stanstarks Please help check if this PR affects BlendMask's performance. Thank you.

@tianzhi0549 tianzhi0549 requested a review from stan-haochen May 25, 2020 11:44
@stan-haochen
Copy link
Collaborator

We should use the displayed (sqrt) score for pre_nms_thresholding. Probably we should modify the default INFERENCE_TH to sqrt(0.05).

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

Successfully merging this pull request may close these issues.

3 participants