Add dim check while calculating mIoU&pixAcc for segmentation test

docs/model_zoo/segmentation.rst

@@ -31,7 +31,7 @@ Table of pre-trained models for semantic segmentation and their performance.
  The test script :download:`Download test.py<../../scripts/segmentation/test.py>` can be used for
  evaluating the models (VOC results are evaluated using the official server). For example ``fcn_resnet50_ade``::
 
- python test.py --dataset ade20k --model-zoo fcn_resnet50_ade --eval
+ python test.py --dataset ade20k --pretrained --model fcn --backbone resnet50 --eval
 
  The training commands work with the script: :download:`Download train.py<../../scripts/segmentation/train.py>`
 

gluoncv/utils/metrics/segmentation.py

@@ -7,15 +7,19 @@
 __all__ = ['SegmentationMetric', 'batch_pix_accuracy', 'batch_intersection_union',
  'pixelAccuracy', 'intersectionAndUnion']
 
+
 class SegmentationMetric(EvalMetric):
  """Computes pixAcc and mIoU metric scores
  """
+
+
  def __init__(self, nclass):
  super(SegmentationMetric, self).__init__('pixAcc & mIoU')
  self.nclass = nclass
  self.lock = threading.Lock()
  self.reset()
 
+
  def update(self, labels, preds):
  """Updates the internal evaluation result.
 
@@ -27,6 +31,8 @@ def update(self, labels, preds):
  preds : 'NDArray' or list of `NDArray`
  Predicted values.
  """
+
+
  def evaluate_worker(self, label, pred):
  correct, labeled = batch_pix_accuracy(
  pred, label)
@@ -38,18 +44,20 @@ def evaluate_worker(self, label, pred):
  self.total_inter += inter
  self.total_union += union
 
+
  if isinstance(preds, mx.nd.NDArray):
  evaluate_worker(self, labels, preds)
  elif isinstance(preds, (list, tuple)):
  threads = [threading.Thread(target=evaluate_worker,
  args=(self, label, pred),
- )
+  )
  for (label, pred) in zip(labels, preds)]
  for thread in threads:
  thread.start()
  for thread in threads:
  thread.join()
 
+
  def get(self):
  """Gets the current evaluation result.
 
@@ -63,13 +71,15 @@ def get(self):
  mIoU = IoU.mean()
  return pixAcc, mIoU
 
+
  def reset(self):
  """Resets the internal evaluation result to initial state."""
  self.total_inter = 0
  self.total_union = 0
  self.total_correct = 0
  self.total_label = 0
 
+
 def batch_pix_accuracy(output, target):
  """PixAcc"""
  # inputs are NDarray, output 4D, target 3D
@@ -79,7 +89,7 @@ def batch_pix_accuracy(output, target):
  target = target.asnumpy().astype('int64') + 1
 
  pixel_labeled = np.sum(target > 0)
- pixel_correct = np.sum((predict == target)*(target > 0))
+ pixel_correct = np.sum((predict == target) * (target > 0))
 
  assert pixel_correct <= pixel_labeled, "Correct area should be smaller than Labeled"
  return pixel_correct, pixel_labeled
@@ -119,7 +129,7 @@ def pixelAccuracy(imPred, imLab):
  # Remove classes from unlabeled pixels in gt image.
  # We should not penalize detections in unlabeled portions of the image.
  pixel_labeled = np.sum(imLab > 0)
- pixel_correct = np.sum((imPred == imLab)*(imLab > 0))
+ pixel_correct = np.sum((imPred == imLab) * (imLab > 0))
  pixel_accuracy = 1.0 * pixel_correct / pixel_labeled
  return (pixel_accuracy, pixel_correct, pixel_labeled)
 

scripts/segmentation/test.py

@@ -120,7 +120,7 @@ def test(model, args, input_transform):
  tbar = tqdm(test_data)
  for i, (data, dsts) in enumerate(tbar):
  if args.eval:
- predicts = [pred[0] for pred in evaluator.parallel_forward(data)]
+ predicts = [pred for pred in evaluator.parallel_forward(data)]
  targets = [target.as_in_context(predicts[0].context) \
  for target in dsts]
  metric.update(targets, predicts)