torch_npu.npu_grid_assign_positive

torch_npu.npu_grid_assign_positive(self, overlaps, box_responsible_flags, max_overlaps, argmax_overlaps, gt_max_overlaps, gt_argmax_overlaps, num_gts, pos_iou_thr, min_pos_iou, gt_max_assign_all) -> Tensor

执行position-sensitive的候选区域池化梯度计算。

• 参数解释：
  • self (Tensor) - float16或float32类型的张量, shape为(n, )。
  • overlaps (Tensor) - 数据类型与assigned_gt_inds相同，表示gt_bboxes和bboxes之间的IoU，shape为(k,n)。
  • box_responsible_flags (Tensor) - 支持uint8数据类型。表示框是否responsible的标志。
  • max_overlaps (Tensor) - 数据类型与assigned_gt_inds. overlaps.max(axis=0)相同。
  • argmax_overlaps (Tensor) - 支持uint32数据类型，overlaps.argmax(axis=0)。
  • gt_max_overlaps (Tensor) - 数据类型与assigned_gt_inds. overlaps.max(axis=1)相同。
  • gt_argmax_overlaps (Tensor) - 支持uint32数据类型， overlaps.argmax(axis=1)。
  • num_gts (Tensor) - 支持uint32数据类型，real k ，shape为 (1, )。
  • pos_iou_thr (Float) - 正检测框的IoU阈值。
  • min_pos_iou (Float) - 检测框被视为正检测框的最小IoU
  • gt_max_assign_all (Bool) - 是否将与某个gt有相同最高重叠的所有检测框分配给该gt。
• 约束条件：

  无

• 示例：

  >>> assigned_gt_inds = torch.rand(4).npu()
  >>> overlaps = torch.rand(2,4).npu()
  >>> box_responsible_flags = torch.tensor([1, 1, 1, 0], dtype=torch.uint8).npu()
  >>> max_overlap = torch.rand(4).npu()
  >>> argmax_overlap = torch.tensor([1, 0, 1, 0], dtype=torch.int32).npu()
  >>> gt_max_overlaps = torch.rand(2).npu()
  >>> gt_argmax_overlaps = torch.tensor([1, 0],dtype=torch.int32).npu()
  >>> output = torch_npu.npu_grid_assign_positive(assigned_gt_inds, overlaps, box_responsible_flags, max_overlap, argmax_overlap, gt_max_overlaps, gt_argmax_overlaps, 128, 0.5, 0., True)
  >>> output.shape
  torch.Size([4])