文档

Upsample(darknet)

Input

x:
  • Required: yes
  • Type: float16, float32
  • Description: y[N, C, i, j] = scale * x[N, C, i/stride_h, j/stride_w], where, 0<=i<=stride_h * x.h – 1, 0 <= j <= stride_w * x.w – 1
  • Restriction: Must be of type float16.

Attribute

  • stride:
    • Required: no
    • Type: int
    • Description: If stride and stride_h/stride_w are both specified, stride applies.
    • Restrictions: none
  • stride_h:
    • Required: no
    • Type: int
    • Description: stride height (default = 2)
    • Restriction: Must be greater than 1.
  • stride_w:
    • Required: no
    • Type: int
    • Description: stride width (default = 2)
    • Restriction: Must be greater than 1.
  • scale:
    • Required: no
    • Type: float
    • Description: scale factor (default = 1)
    • Restrictions: none

Output

y:
  • Required: yes
  • Type: float16, float32
  • Description: y[N, C, i, j] = scale * x[N, C, i/stride_h, j/stride_w], where, 0<=i<=stride_h * x.h – 1, 0 <= j <= stride_w * x.w – 1
  • Restriction: Must be of type float16.