`mlvbench.metrics`¶

Evaluation metrics.

`intersection_over_union(prediction, target)` ¶

Compute the intersection over union (IoU) of two binary masks.

\[ \mathrm{IoU} = \frac{|P \cap T|}{|P \cup T|} \]

If both the prediction and target are empty (\(|P \cup T| = 0\)), this returns 1.0 since the prediction is correct.

Parameters:

Name	Type	Description	Default
`prediction`	`Tensor`	The predicted mask, a `bool` tensor of arbitrary shape.	required
`target`	`Tensor`	The target mask, a `bool` tensor of the same shape as `prediction`.	required

Returns:

Type	Description
`float`	The IoU computed over all elements. Returns 1.0 if both masks are empty.