add F1 metrics

k3_addons/metrics/f_scores.py

@@ -1,6 +1,7 @@
 from keras import ops, metrics
+from k3_addons.api_export import k3_export
 
-
+@k3_export(path="k3_addons.metrics.FBetaScore")
 class FBetaScore(metrics.Metric):
  def __init__(
  self,
@@ -123,3 +124,20 @@ def reset_state(self):
  reset_value = ops.zeros(self.init_shape, dtype=self.dtype)
  for v in self.variables:
  v.assign(reset_value)
+
+@k3_export(path="k3_addons.metrics.F1Score")
+class F1Score(FBetaScore):
+ def __init__(
+ self,
+ num_classes,
+ average=None,
+ threshold=None,
+ name="f1_score",
+ dtype=None,
+ ):
+ super().__init__(num_classes, average, 1.0, threshold, name=name, dtype=dtype)
+
+ def get_config(self):
+ base_config = super().get_config()
+ del base_config["beta"]
+ return base_config

k3_addons/metrics/f_scores_test.py

@@ -2,7 +2,7 @@
 
 import pytest
 
-from k3_addons.metrics.f_scores import FBetaScore
+from k3_addons.metrics.f_scores import FBetaScore, F1Score
 from k3_addons.metrics.utils import _get_model
 from keras import ops, backend
 
@@ -167,3 +167,55 @@ def test_fbeta_weighted_random_score_none(avg_val, beta, sample_weights, result)
 def test_keras_model():
  fbeta = FBetaScore(5, "micro", 1.0)
  _get_model(fbeta, 5)
+
+
+def test_eq():
+ f1 = F1Score(3)
+ fbeta = FBetaScore(3, beta=1.0)
+
+ preds = [
+ [0.9, 0.1, 0],
+ [0.2, 0.6, 0.2],
+ [0, 0, 1],
+ [0.4, 0.3, 0.3],
+ [0, 0.9, 0.1],
+ [0, 0, 1],
+ ]
+ actuals = [[1, 0, 0], [0, 1, 0], [0, 0, 1], [1, 0, 0], [1, 0, 0], [0, 0, 1]]
+ preds = ops.convert_to_tensor(preds, "float32")
+ actuals = ops.convert_to_tensor(actuals, "float32")
+ fbeta.update_state(actuals, preds)
+ f1.update_state(actuals, preds)
+ np.testing.assert_allclose(fbeta.result().numpy(), f1.result().numpy())
+
+
+def test_sample_eq():
+ f1 = F1Score(3)
+ f1_weighted = F1Score(3)
+
+ preds = ops.convert_to_tensor([
+ [0.9, 0.1, 0],
+ [0.2, 0.6, 0.2],
+ [0, 0, 1],
+ [0.4, 0.3, 0.3],
+ [0, 0.9, 0.1],
+ [0, 0, 1],
+ ])
+ actuals = ops.convert_to_tensor([[1, 0, 0], [0, 1, 0], [0, 0, 1], [1, 0, 0], [1, 0, 0], [0, 0, 1]])
+ sample_weights = [1.0, 1.0, 1.0, 1.0, 1.0, 1.0]
+
+ f1.update_state(actuals, preds)
+ f1_weighted(actuals, preds, sample_weights)
+ np.testing.assert_allclose(f1.result().numpy(), f1_weighted.result().numpy())
+
+
+def test_keras_model_f1():
+ f1 = F1Score(5)
+ _get_model(f1, 5)
+
+
+def test_config_f1():
+ f1 = F1Score(3)
+ config = f1.get_config()
+ assert "beta" not in config
+