semantic_kitti_metric.py support to print best acc/iou (PaddlePaddle#391

)
kongdebug · Jul 19, 2023 · e5d0dc6 · e5d0dc6
1 parent fe09257
commit e5d0dc6
Show file tree

Hide file tree

Showing 2 changed files with 14 additions and 3 deletions.
diff --git a/paddle3d/apis/trainer.py b/paddle3d/apis/trainer.py
@@ -141,6 +141,7 @@ def __init__(
         self.eval_dataloader = _dataloader_build_fn(
             val_dataset, self.model) if val_dataset else None
         self.val_dataset = val_dataset
+        self.eval_metric_obj = self.val_dataset.metric
 
         self.profiler_options = profiler_options
         self.resume = resume
@@ -471,11 +472,11 @@ def evaluate(self) -> float:
         if self.val_dataset is None:
             raise RuntimeError('No evaluation dataset specified!')
         msg = 'evaluate on validate dataset'
-        metric_obj = self.val_dataset.metric
 
         for idx, sample in self.logger.enumerate(self.eval_dataloader, msg=msg):
             result = validation_step(self.model, sample)
-            metric_obj.update(predictions=result, ground_truths=sample)
+            self.eval_metric_obj.update(
+                predictions=result, ground_truths=sample)
 
-        metrics = metric_obj.compute(verbose=True)
+        metrics = self.eval_metric_obj.compute(verbose=True)
         return metrics
diff --git a/paddle3d/datasets/semantic_kitti/semantic_kitti_metric.py b/paddle3d/datasets/semantic_kitti/semantic_kitti_metric.py
@@ -51,6 +51,9 @@ def __init__(self, num_classes: int, ignore: List[int] = None):
             [n for n in range(self.num_classes) if n not in ignore],
             dtype="int64")
 
+        self.best_acc_avg = 0
+        self.best_iou_avg = 0
+
         # reset the class counters
         self.reset()
 
@@ -115,6 +118,10 @@ def getacc(self):
     def compute(self, verbose=False) -> dict:
         m_accuracy = self.getacc()
         m_jaccard, class_jaccard = self.getIoU()
+        if m_accuracy > self.best_acc_avg:
+            self.best_acc_avg = m_accuracy
+        if m_jaccard > self.best_iou_avg:
+            self.best_iou_avg = m_jaccard
 
         if verbose:
             logger.info("Acc avg {:.3f}".format(float(m_accuracy)))
@@ -129,5 +136,8 @@ def compute(self, verbose=False) -> dict:
                                 SemanticKITTIDataset.LEARNING_MAP_INV[i]],
                             jacc=float(jacc)))
 
+            logger.info("Best Acc avg {:.3f}".format(float(self.best_acc_avg)))
+            logger.info("Best IoU avg {:.3f}".format(float(self.best_iou_avg)))
+
         return dict(
             mean_acc=m_accuracy, mean_iou=m_jaccard, class_iou=class_jaccard)