torch_npu.npu_nms_v4

torch_npu.npu_nms_v4(boxes, scores, max_output_size, iou_threshold, scores_threshold, pad_to_max_output_size=False) -> (Tensor, Tensor)

按分数降序选择标注框的子集。