ROIAlign
Input
- features:
- Required: yes
- Type: float16, float32
- Description: input tensor
- Restrictions: none
- rois:
- Required: yes
- Type: float16, float32
- Description: ROI information with shape [N, 5-tuple] or [N, 8-tuple]. The 5-tuple is formatted as (batchID, x1, y1, x2, y2). For the 8-tuple, only the first 5 tuples are valid.
- Restrictions: none
- rois_n:
- Required: no
- Type: int32
- Description: actual number of ROIs per batch. Has shape [batch, 8]. The value 8 indicates that only the first column in the eight columns is valid (data padding is performed for performance acceleration, and the rest seven columns are invalid data that is padded).
For example, for shape [4, 8], only the first column is valid, indicating the number of ROIs per batch.
[0,0]...[0,7]
[1,0]...[1,7]
[2,0]...[2,7]
[3,0]...[3,7]
- Restriction: This parameter is optional.
Attribute
- spatial_scale:
- Required: yes
- Type: float
- Description: multiplicative spatial scale factor to translate ROI coordinates from their input scale to the scale used when pooling.
- Restrictions: none
- pooled_h:
- Required: yes
- Type: int
- Description: height of output y
- Restrictions: none
- pooled_w:
- Required: yes
- Type: int
- Description: width of output y
- Restrictions: none
- sampling_ratio:
- Required: no
- Type: int
- Description: used to determine the divided bins
- Restriction: Must be 2.
Output
output:
- Required: yes
- Type: float16, float32
- Description: none
- Restrictions: none
Parent topic: Supported Caffe Operators