Add ESC-50 evaluation results via explicit mapping

model_checkpoints/Primitive_Extractor/FSD50k/evaluation_result_of_part2_tested_on_esc50.json

@@ -0,0 +1,15 @@
+{
+  "dataset": "ESC-50 (Full, including Fold 5)",
+  "evaluation_method": "Explicit mapping ESC-50 \u2192 FSD50K",
+  "total_samples_evaluated": 1040,
+  "esc50_classes_mapped": 26,
+  "fsd50k_classes_detected": 199,
+  "metrics": {
+    "mAP_macro_percent": 10.14,
+    "micro_f1_score_percent": 18.69,
+    "macro_f1_score_percent": 7.6,
+    "top_1_accuracy_percent": 18.37
+  },
+  "checkpoint_used": "best_fsd50k_primitive_part1_and_2.pt",
+  "analysis": "Cross-dataset evaluation via explicit class mapping. Top-1 Accuracy of 18.37% demonstrates how well the FSD50K-trained Primitive Extractor generalizes to ESC-50 environmental sounds."
+}