roi_align¶

dragon.vm.torchvision.ops.roi_align(
  input,
  boxes,
  output_size,
  spatial_scale=1.0,
  sampling_ratio=- 1,
  aligned=False
)[source]¶

Apply the average roi align to input. [He et.al, 2017].

The boxes should be packed with a shape like \((N, 5)\), where \(N\) is the number of boxes.

Each box is a 5d sequence containing (batch_index, x1, y1, x2, y2).

Parameters:

• input (dragon.vm.torch.Tensor) – The input tensor.
• boxes (dragon.vm.torch.Tensor) – The box coordinates.
• output_size (Sequence[int]) – The output height and width.
• spatial_scale (float, optional, default=1.0) – The input scale to the size of boxes.
• sampling_ratio (int, optional, default=-1) – The number of sampling grids for boxes.
• aligned (bool, optional, default=False) – Whether to shift the input coordinates by -0.5.

Returns:

dragon.vm.torch.Tensor – The output tensor.