Skip detr: end-to-end skip connection model for small object detection in forestry pest dataset

HIGHLIGHTS

• who: DETR and colleagues from the Manchester Metropolitan University, United Kingdom have published the Article: Skip DETR: end-to-end Skip connection model for small object detection in forestry pest dataset, in the Journal: (JOURNAL)
• what: At the same time, compared with the initial DETR model, the model has achieved competitive results. The authors propose a model called Skip DETR, which uses skip connection to enhance the extraction of features of small sample images by the DETR model. The authors design a three-layer spatial pyramid pooling layer, decompose the feature map into 2u00d72, 4u00d74 . . .
  

   

  

  If you want to have access to all the content you need to log in!

  Thanks :)

  Username or Email

  Password

  Remember me

  Lost your password?

  If you don't have an account, you can create one here.