A lightweight multi-dimension dynamic convolutional network for real-time semantic segmentation

HIGHLIGHTS

• who: Chunyu Zhang from the Yunnan University, China have published the Article: A lightweight multi-dimension dynamic convolutional network for real-time semantic segmentation, in the Journal: (JOURNAL)
• what: In practical application scenarios, slow inference speed and many parameters are the main reasons semantic segmentation cannot be applied. The authors design a new multi-dimensional dynamic convolution (MDy-Conv), which uses an attention mechanism for convolution and linearly combines multiple factorial convolutions to find a convolution kernel suitable for the current feature. The authors design the MS-DAB module to include MDyConv, residual connections, and . . .
  

   

  

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

  Thanks :)

  Username or E-mail:

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