Memory Leak in BinarPrecisionRecallCurve

[suahelen]

Hi Folks,

I stumbled over this issue while playing around with anomlib

🐛 Bug

See this issue in anomalib:
openvinotoolkit/anomalib#2531

To keep it short continuously calling compute on BinaryPrecisionRecallCurve(Metric) grows the self.preds and self.target indefinitely.

This eventually leads to a memory overflow.

the following code resolves the issue by clearing the preds and targets after concatenating them to the state

    def compute(self) -> tuple[Tensor, Tensor, Tensor]:
        """Compute metric."""
        if self.thresholds is None:
            if not self.preds or not self.target:
                return torch.tensor([]), torch.tensor([]), torch.tensor([])
            state = (torch.cat(self.preds), torch.cat(self.target))

            self.preds.clear()
            self.target.clear()
        else:
            state = self.confmat
            self.confmat.zero_()

        precision, recall, thresholds = _binary_precision_recall_curve_compute(state, self.thresholds)
        return precision, recall, thresholds if thresholds is not None else torch.tensor([])

Cheers

[github-actions]

Hi! thanks for your contribution!, great first issue!

[rittik9]

@Borda I would be interested in working on this