Fix ignore_index in jaccard index

CHANGELOG.md

@@ -90,6 +90,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Fixed classification metrics for `byte` input ([#1521](https://github.com/Lightning-AI/metrics/pull/1474))
 
 
+- Fixed the use of `ignore_index` in `MulticlassJaccardIndex` ([#1386](https://github.com/Lightning-AI/metrics/pull/1386))
+
+
 ## [0.11.1] - 2023-01-30
 
 ### Fixed

src/torchmetrics/classification/jaccard.py

@@ -176,7 +176,7 @@ def __init__(
 
     def compute(self) -> Tensor:
         """Compute metric."""
-        return _jaccard_index_reduce(self.confmat, average=self.average)
+        return _jaccard_index_reduce(self.confmat, average=self.average, ignore_index=self.ignore_index)
 
 
 class MultilabelJaccardIndex(MultilabelConfusionMatrix):

src/torchmetrics/functional/classification/jaccard.py

@@ -38,6 +38,7 @@
 def _jaccard_index_reduce(
     confmat: Tensor,
     average: Optional[Literal["micro", "macro", "weighted", "none", "binary"]],
+    ignore_index: Optional[int] = None,
 ) -> Tensor:
     """Perform reduction of an un-normalized confusion matrix into jaccard score.
 
@@ -53,6 +54,9 @@ def _jaccard_index_reduce(
               metrics across classes, weighting each class by its support (``tp + fn``).
             - ``'none'`` or ``None``: Calculate the metric for each class separately, and return
               the metric for every class.
+
+        ignore_index:
+            Specifies a target value that is ignored and does not contribute to the metric calculation
     """
     allowed_average = ["binary", "micro", "macro", "weighted", "none", None]
     if average not in allowed_average:
@@ -61,6 +65,7 @@ def _jaccard_index_reduce(
     if average == "binary":
         return confmat[1, 1] / (confmat[0, 1] + confmat[1, 0] + confmat[1, 1])
     else:
+        ignore_index_cond = ignore_index is not None and 0 <= ignore_index <= confmat.shape[0]
         if confmat.ndim == 3:  # multilabel
             num = confmat[:, 1, 1]
             denom = confmat[:, 1, 1] + confmat[:, 0, 1] + confmat[:, 1, 0]
@@ -70,16 +75,18 @@ def _jaccard_index_reduce(
 
         if average == "micro":
             num = num.sum()
-            denom = denom.sum()
+            denom = denom.sum() - (denom[ignore_index] if ignore_index_cond else 0.0)
 
         jaccard = _safe_divide(num, denom)
 
-        if average is None or average == "none":
+        if average is None or average == "none" or average == "micro":
             return jaccard
         if average == "weighted":
             weights = confmat[:, 1, 1] + confmat[:, 1, 0] if confmat.ndim == 3 else confmat.sum(1)
         else:
             weights = torch.ones_like(jaccard)
+            if ignore_index_cond:
+                weights[ignore_index] = 0.0
         return ((weights * jaccard) / weights.sum()).sum()
 
 
@@ -217,7 +224,7 @@ def multiclass_jaccard_index(
         _multiclass_confusion_matrix_tensor_validation(preds, target, num_classes, ignore_index)
     preds, target = _multiclass_confusion_matrix_format(preds, target, ignore_index)
     confmat = _multiclass_confusion_matrix_update(preds, target, num_classes)
-    return _jaccard_index_reduce(confmat, average=average)
+    return _jaccard_index_reduce(confmat, average=average, ignore_index=ignore_index)
 
 
 def _multilabel_jaccard_index_arg_validation(
@@ -297,7 +304,7 @@ def multilabel_jaccard_index(
         _multilabel_confusion_matrix_tensor_validation(preds, target, num_labels, ignore_index)
     preds, target = _multilabel_confusion_matrix_format(preds, target, num_labels, threshold, ignore_index)
     confmat = _multilabel_confusion_matrix_update(preds, target, num_labels)
-    return _jaccard_index_reduce(confmat, average=average)
+    return _jaccard_index_reduce(confmat, average=average, ignore_index=ignore_index)
 
 
 def jaccard_index(

tests/unittests/classification/test_jaccard.py

@@ -124,6 +124,10 @@ def _sklearn_jaccard_index_multiclass(preds, target, ignore_index=None, average=
     preds = preds.flatten()
     target = target.flatten()
     target, preds = remove_ignore_index(target, preds, ignore_index)
+    if ignore_index is not None and 0 <= ignore_index <= NUM_CLASSES:
+        labels = [i for i in range(NUM_CLASSES) if i != ignore_index]
+        res = sk_jaccard_index(y_true=target, y_pred=preds, average=average, labels=labels)
+        return np.insert(res, ignore_index, 0.0) if average is None else res
     return sk_jaccard_index(y_true=target, y_pred=preds, average=average)
 
 
@@ -236,7 +240,7 @@ def _sklearn_jaccard_index_multilabel(preds, target, ignore_index=None, average=
 @pytest.mark.parametrize("input", _multilabel_cases)
 class TestMultilabelJaccardIndex(MetricTester):
     @pytest.mark.parametrize("average", ["macro", "micro", "weighted", None])
-    @pytest.mark.parametrize("ignore_index", [None])  # , -1, 0])
+    @pytest.mark.parametrize("ignore_index", [None, -1])
     @pytest.mark.parametrize("ddp", [True, False])
     def test_multilabel_jaccard_index(self, input, ddp, ignore_index, average):
         preds, target = input
@@ -256,7 +260,7 @@ def test_multilabel_jaccard_index(self, input, ddp, ignore_index, average):
         )
 
     @pytest.mark.parametrize("average", ["macro", "micro", "weighted", None])
-    @pytest.mark.parametrize("ignore_index", [None, -1, 0])
+    @pytest.mark.parametrize("ignore_index", [None, -1])
     def test_multilabel_jaccard_index_functional(self, input, ignore_index, average):
         preds, target = input
         if ignore_index is not None: