mindspore.ops.iou

mindspore.ops.iou(anchor_boxes, gt_boxes, mode='iou')

Calculates intersection over union for boxes.

Computes the intersection over union (IOU) or the intersection over foreground (IOF) based on the ground-truth and predicted regions.

$$\begin{array}{r}\begin{array}{c}\text{IOU}=\frac{\text{Area of Overlap}}{\text{Area of Union}}\\ \text{IOF}=\frac{\text{Area of Overlap}}{\text{Area of Ground Truth}}\end{array}\end{array}$$

Warning

In Ascend, only computation of float16 data is supported. To avoid overflow, the input length and width are scaled by 0.2 internally.

Parameters

• anchor_boxes (Tensor) – Anchor boxes, tensor of shape $(N,4)$ . $N$ indicates the number of anchor boxes, and the value $4$ refers to four boundary coordinates of the predicted area "x0", "y0", "x1", and "y1". Data type must be either float16, float32 or float64.

• gt_boxes (Tensor) – Ground truth boxes, tensor of shape $(M,4)$ . $M$ indicates the number of ground truth boxes, and the value $4$ refers to four boundary coordinates of the truth area "x0", "y0", "x1", and "y1". Data type must be either float16, float32 or float64.

• mode (string) – The mode is used to specify the calculation method, now supporting 'iou' (intersection over union) or 'iof' (intersection over foreground) mode. Default: 'iou' .

Returns

Tensor, the IOU/IOF values, tensor of shape $(M,N)$ , with the same data type as anchor_boxes.

Raises

KeyError – When mode is not 'iou' or 'iof'.

Supported Platforms:

Ascend GPU CPU

Examples

>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> anchor_boxes = Tensor(np.random.randint(1.0, 5.0, [3, 4]), mindspore.float16)
>>> gt_boxes = Tensor(np.random.randint(1.0, 5.0, [3, 4]), mindspore.float16)
>>> mode = 'iou'
>>> output = ops.iou(anchor_boxes, gt_boxes, mode)
>>> print(output.shape)
(3, 3)