diff --git a/perceptionmetrics/utils/detection_metrics.py b/perceptionmetrics/utils/detection_metrics.py
index e6ac4f10..c659fbba 100644
--- a/perceptionmetrics/utils/detection_metrics.py
+++ b/perceptionmetrics/utils/detection_metrics.py
@@ -182,10 +182,20 @@ def compute_metrics(self) -> Dict[int, Dict[str, float]]:
 
             ap, precision, recall = compute_ap(tps, fps, fn_count)
 
+            precision_val = precision[-1] if len(precision) > 0 else 0
+            recall_val = recall[-1] if len(recall) > 0 else 0
+
+            f1_score = (
+                2 * precision_val * recall_val / (precision_val + recall_val)
+                if (precision_val + recall_val) > 0
+                else 0.0
+            )
+
             metrics[label] = {
                 "AP": ap,
-                "Precision": precision[-1] if len(precision) > 0 else 0,
-                "Recall": recall[-1] if len(recall) > 0 else 0,
+                "Precision": precision_val,
+                "Recall": recall_val,
+                "F1": f1_score,
                 "TP": sum(tps),
                 "FP": sum(fps),
                 "FN": fn_count,
@@ -301,7 +311,7 @@ def get_overall_precision_recall_curve(self) -> Dict[str, List[float]]:
             return {"precision": [0.0], "recall": [0.0]}
 
         fn_count = sum(1 for d in all_detections if d[1] == -1)
-        
+
         # Sort by score
         all_detections = sorted(
             [d for d in all_detections if d[0] is not None], key=lambda x: -x[0]
@@ -356,7 +366,7 @@ def get_metrics_dataframe(self, ontology: dict) -> pd.DataFrame:
         metrics_dict = {}
         class_names = list(ontology.keys())
 
-        for metric in ["AP", "Precision", "Recall", "TP", "FP", "FN"]:
+        for metric in ["AP", "Precision", "Recall", "F1", "TP", "FP", "FN"]:
             metrics_dict[metric] = {}
             for class_name, class_data in ontology.items():
                 idx = class_data["idx"]
diff --git a/tests/test_detection_metrics.py b/tests/test_detection_metrics.py
index a06f3ef0..31bbc38d 100644
--- a/tests/test_detection_metrics.py
+++ b/tests/test_detection_metrics.py
@@ -32,9 +32,10 @@ def test_match_predictions_logic(metrics_factory):
     results = matches[1]
     assert (0.95, 1) in results, "High overlap prediction should be a True Positive"
     assert (0.6, 0) in results, "Zero overlap prediction should be a False Positive"
-    assert (None, -1) not in results, (
-        "GT was matched, so there should be no False Negative"
-    )
+    assert (
+        None,
+        -1,
+    ) not in results, "GT was matched, so there should be no False Negative"
 
 
 def test_compute_metrics(metrics_factory):
@@ -138,9 +139,9 @@ def test_compute_coco_map_perfect_match(metrics_factory):
     metrics_factory.update(gt_boxes, [1], pred_boxes, [1], [0.9])
     coco_map = metrics_factory.compute_coco_map()
 
-    assert coco_map == 1.0, (
-        "Perfect overlap must yield perfect mAP across all thresholds"
-    )
+    assert (
+        coco_map == 1.0
+    ), "Perfect overlap must yield perfect mAP across all thresholds"
 
 
 def test_compute_coco_map_complex_multi_class(metrics_factory):
@@ -232,3 +233,34 @@ def test_coco_map_empty_vs_non_empty_class(metrics_factory):
 
     # (AP_Class1 + AP_Class2) / 2  => (1.0 + 0.0) / 2 = 0.5
     assert np.isclose(mAP, 0.5), f"Expected mAP of 0.5, but got {mAP}"
+
+
+def test_f1_score(metrics_factory):
+    """
+    Verify F1 score calculation.
+    """
+
+    gt_boxes = np.array([[10, 10, 50, 50]])
+    gt_labels = [1]
+
+    pred_boxes = np.array(
+        [
+            [12, 12, 48, 48],  # TP
+            [11, 11, 49, 49],  # FP (duplicate detection)
+        ]
+    )
+
+    pred_labels = [1, 1]
+    pred_scores = [0.9, 0.4]
+
+    metrics_factory.update(
+        gt_boxes,
+        gt_labels,
+        pred_boxes,
+        pred_labels,
+        pred_scores,
+    )
+
+    metrics = metrics_factory.compute_metrics()
+
+    assert np.isclose(metrics[1]["F1"], 2 / 3)