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

DSSD : Deconvolutional Single Shot Detector #62

Open
chullhwan-song opened this issue Jan 21, 2019 · 2 comments
Open

DSSD : Deconvolutional Single Shot Detector #62

chullhwan-song opened this issue Jan 21, 2019 · 2 comments

Comments

@chullhwan-song
Copy link
Owner

https://arxiv.org/abs/1701.06659

@chullhwan-song
Copy link
Owner Author

chullhwan-song commented Jun 28, 2019

Abstract

  • 기본적으로 SSD 알고리즘 Based
  • 여기서 adding하는 부분은
    • base network + Deconvolution layer에 대한 영역. = DSSD
    • Deconvolution 은 U-Net으로 봐도 무방할듯~
  • 간단리뷰

SSD vs DSSD

image

  • basenet 중간부터 U-Net 형태를 구성한다.
  • SSD >VGG vs DSS > ResNet101
  • Prediction layer
    image
    • SSD는 Fig.2 a에서 부터 직접 변형?이 안된 그대로의 feature map를 가져다가 class/bos regression를 이용했지만
    • DSSD는 그 feature map를 이용하여 다시 advance(?)한 여러 conv를 이용하여 feature map를 구성함
  • Deconvolution 구성.
    image
    • U-Net과 거의 유사
    • 다른점은 concat하는 형태로 결합하는 구조였는데, element wise 형태로 결합
      • refinement network

@chullhwan-song
Copy link
Owner Author

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

No branches or pull requests

1 participant