Metrics

Classification & Segmentation

Accuracy

common.utils.metric.accuracy(output, target, topk=(1, ))

Computes the accuracy over the k top predictions for the specified values of k

Parameters

• output (tensor) – Classification outputs, \((N, C)\) where C = number of classes

• target (tensor) – \((N)\) where each value is \(0 \leq \text{targets}[i] \leq C-1\)

• topk (sequence[int]) – A list of top-N number.

Returns

Top-N accuracies (N \(\in\) topK).

ConfusionMatrix

class common.utils.metric.ConfusionMatrix(num_classes)

compute()

compute global accuracy, per-class accuracy and per-class IoU

format(classes)

Get the accuracy and IoU for each class in the table format

update(target, output)

Update confusion matrix.

Parameters

• target – ground truth

• output – predictions of models

Shape:

• target: \((minibatch, C)\) where C means the number of classes.

• output: \((minibatch, C)\) where C means the number of classes.