For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Mojo function

roi_align_shape

def roi_align_shape(input: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=input.static_spec], rois: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=rois.static_spec], output_height: Int64, output_width: Int64, spatial_scale: Scalar, sampling_ratio: Scalar) -> IndexList[Int(4)]

Computes the output shape for the mo.roi_align graph op.

Args:

• ​input (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=input.static_spec]): Rank-4 batched image input in NHWC format with shape (N, H, W, C).
• ​rois (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=rois.static_spec]): Rank-2 tensor of ROI box coordinates with shape (M, 5) where each row is (batch_idx, y0, x0, y1, x1).
• ​output_height (Int64): Pooled output height, in elements.
• ​output_width (Int64): Pooled output width, in elements.
• ​spatial_scale (Scalar): Scale factor remapping ROI coordinates to input coordinates.
• ​sampling_ratio (Scalar): Number of sampling points in the interpolation grid used to compute each pooled bin.

Returns:

IndexList[Int(4)]: The output shape (num_rois, output_height, output_width, channels).