openvino.runtime.opset9.multiclass_nms#

openvino.runtime.opset9.multiclass_nms(boxes: Node | int | float | ndarray, scores: Node | int | float | ndarray, roisnum: Node | int | float | ndarray | None = None, sort_result_type: str | None = 'none', sort_result_across_batch: bool | None = False, output_type: str | None = 'i64', iou_threshold: float | None = 0.0, score_threshold: float | None = 0.0, nms_top_k: int | None = -1, keep_top_k: int | None = -1, background_class: int | None = -1, nms_eta: float | None = 1.0, normalized: bool | None = True, name: str | None = None) → Node#

Return a node which performs MulticlassNms.

Parameters:

  • boxes – Tensor with box coordinates.

  • scores – Tensor with box scores.

  • roisnum – Tensor with roisnum. Specifies the number of rois in each image. Required when ‘scores’ is a 2-dimensional tensor.

  • sort_result_type – Specifies order of output elements, possible values: ‘class’: sort selected boxes by class id (ascending) ‘score’: sort selected boxes by score (descending) ‘none’: do not guarantee the order.

  • sort_result_across_batch – Specifies whenever it is necessary to sort selected boxes across batches or not

  • output_type – Specifies the output tensor type, possible values: ‘i64’, ‘i32’

  • iou_threshold – Specifies intersection over union threshold

  • score_threshold – Specifies minimum score to consider box for the processing

  • nms_top_k – Specifies maximum number of boxes to be selected per class, -1 meaning to keep all boxes

  • keep_top_k – Specifies maximum number of boxes to be selected per batch element, -1 meaning to keep all boxes

  • background_class – Specifies the background class id, -1 meaning to keep all classes

  • nms_eta – Specifies eta parameter for adpative NMS, in close range [0, 1.0]

  • normalized – Specifies whether boxes are normalized or not

  • name – The optional name for the output node

Returns:

The new node which performs MuticlassNms