megengine.functional.nn.roi_pooling¶

roi_pooling(inp, rois, output_shape, mode='max', scale=1.0)[源代码]¶

Applies roi pooling on input feature.

参数

inp (Tensor) – tensor that represents the input feature, (N, C, H, W) images.
rois (Tensor) – K, 5)` boxes. First column is the index into N. The other 4 columns are xyxy.
output_shape (Union[int, tuple, list]) – height, width)` of output rois feature.
mode (str) – max” or “average”, use max/average align just like max/average pooling. Default: “max”
scale (float) – scale the input boxes by this number. Default: 1.0

返回类型

Tensor

返回

K, C, output_shape[0], output_shape[1]) feature of rois.

实际案例

import numpy as np
from megengine import tensor
import megengine.functional as F

np.random.seed(42)
inp = tensor(np.random.randn(1, 1, 128, 128))
rois = tensor(np.random.random((4, 5)))
y = F.vision.roi_pooling(inp, rois, (2, 2))
print(y.numpy()[0].round(decimals=4))

Outputs:

[[[-0.1383 -0.1383]
  [-0.5035 -0.5035]]]

megengine.functional.nn.warp_perspective

megengine.functional.nn.roi_align