Error at running Object Segmentation (APOC)

[aloboa]

(running on Pop OS (essentially the same as Ubuntu 22.04 LTS))
Following https://biapol.github.io/BioImage-Analysis-and-Data-Processing-Workshop-2025/interactive_pixel_classification/readme.html,
I get to:

and after I click on train, I get I get an error: dGetPlatformIDs failed
Traceback message:

dGetPlatformIDs failed

---------------------------------------------------------------------------
LogicError                                Traceback (most recent call last)
File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/napari_accelerated_pixel_and_object_classification/_dock_widget.py:161, in ObjectSegmentation.__init__.<locals>.train_clicked(*arg=(False,), **kwargs={})
    157     return
    159 first_image_layer = self.get_selected_images()[0]
--> 161 self.train(
        self = <napari_accelerated_pixel_and_object_classification._dock_widget.ObjectSegmentation object at 0x7db13931b520>
        num_object_annotation_spinner = <PyQt5.QtWidgets.QSpinBox object at 0x7db1391d44c0>
        self.feature_selector = <napari_accelerated_pixel_and_object_classification._dock_widget.FeatureSelector object at 0x7db1391d49d0>
        num_max_depth_spinner = <PyQt5.QtWidgets.QSpinBox object at 0x7db1391e0310>
        num_trees_spinner = <PyQt5.QtWidgets.QSpinBox object at 0x7db1391e03a0>
        filename_edit = FileEdit(mode='w', value=PosixPath('ObjectSegmenter.cl'))
        show_classifier_statistics_checkbox = <PyQt5.QtWidgets.QCheckBox object at 0x7db1391e0670>
        first_image_layer = <Image layer 'blobs' at 0x7db1391f2140>
    162     self.get_selected_images_data(),
    163     self.get_selected_annotation_data(),
    164     num_object_annotation_spinner.value(),
    165     self.feature_selector.getFeatures(),
    166     num_max_depth_spinner.value(),
    167     num_trees_spinner.value(),
    168     str(filename_edit.value.absolute()).replace("\\", "/").replace("//", "/"),
    169     show_classifier_statistics_checkbox.isChecked(),
    170     first_image_layer.scale
    171 )

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/napari_accelerated_pixel_and_object_classification/_dock_widget.py:288, in ObjectSegmentation.train(self=<napari_accelerated_pixel_and_object_classification._dock_widget.ObjectSegmentation object>, images=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8), annotation=array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, ...., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8), object_annotation_value=2, feature_definition='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', num_max_depth=2, num_trees=100, filename='/home/alobo/BioImage-Analysis-and-Data-Processing-Workshop-2025/ObjectSegmenter.cl', show_classifier_statistics=False, scale=array([1., 1.]))
    284     classifier.output_probability_of_class = object_annotation_value
    286 print("annotation shape", annotation.shape)
--> 288 classifier.train(feature_definition, annotation, images)
        classifier = Classifier type: ObjectSegmenter
--- Random forest info ---
Used features for training: gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1
Ground truth dimensions: 2
Maximum depth: 2
Number of ensembles: 100
Number of classes: 
Number of features: 0
Number of channels: 0Positive class identifier: 2
        feature_definition = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        annotation = array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, 0],
       ...,
       [0, 0, ..., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8)
        images = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
    290 print("Training done. Applying model...")
    292 result = np.asarray(classifier.predict(features=feature_definition, image=images))

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_object_segmenter.py:38, in ObjectSegmenter.train(self=Classifier type: ObjectSegmenter
--- Random fore...Number of channels: 0Positive class identifier: 2, features='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', ground_truth=array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, ...., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8), image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8), continue_training=False)
     31 """Train the classifier with.
     32 
     33 See Also
     34 --------
     35 .. PixelClassifier.train()
     36 """
     37 self.positive_class_identifier_from_file = self.positive_class_identifier
---> 38 super().train(features, ground_truth, image, continue_training=continue_training)
        features = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        ground_truth = array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, 0],
       ...,
       [0, 0, ..., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8)
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
        continue_training = False

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_pixel_classifier.py:101, in PixelClassifier.train(self=Classifier type: ObjectSegmenter
--- Random fore...Number of channels: 0Positive class identifier: 2, features='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', ground_truth=array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, ...., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8), image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8), continue_training=False)
     99 # make features and convert in the right format
    100 self.num_ground_truth_dimensions = len(ground_truth.shape)
--> 101 features = self._make_features_potentially_multichannel(features, image)
        features = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        self = Classifier type: ObjectSegmenter
--- Random forest info ---
Used features for training: gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1
Ground truth dimensions: 2
Maximum depth: 2
Number of ensembles: 100
Number of classes: 
Number of features: 0
Number of channels: 0Positive class identifier: 2
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
    102 self.num_features = len(features)
    103 X, y = self._to_np(features, ground_truth)

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_pixel_classifier.py:282, in PixelClassifier._make_features_potentially_multichannel(self=Classifier type: ObjectSegmenter
--- Random fore...Number of channels: 0Positive class identifier: 2, features='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8))
    280     features = new_features
    281 else:
--> 282     features = self._make_features(features, image)
        features = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
        self = Classifier type: ObjectSegmenter
--- Random forest info ---
Used features for training: gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1
Ground truth dimensions: 2
Maximum depth: 2
Number of ensembles: 100
Number of classes: 
Number of features: 0
Number of channels: 0Positive class identifier: 2
    283 return features

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_pixel_classifier.py:304, in PixelClassifier._make_features(self=Classifier type: ObjectSegmenter
--- Random fore...Number of channels: 0Positive class identifier: 2, features='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8))
    302     if image is None:
    303         raise TypeError("If features are provided as string, an image must be given as well to generate features.")
--> 304     features = generate_feature_stack(image, features)
        features = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
    306 return features

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_utils.py:27, in generate_feature_stack(image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8), features_specification='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1')
      9 def generate_feature_stack(image, features_specification : Union[str, PredefinedFeatureSet] = None):
     10     """
     11     Creates a feature stack from a given image.
     12 
   (...)
     24     a list of OCLarray images
     25     """
---> 27     image = cle.push(image)
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
        cle = <module 'pyclesperanto_prototype' from '/home/alobo/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/__init__.py'>
     29     # default features
     30     if features_specification is None:

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_push.py:41, in push(any_array=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8))
     38 if hasattr(any_array, 'shape') and hasattr(any_array, 'dtype') and hasattr(any_array, 'get'):
     39     any_array = np.asarray(any_array.get())
---> 41 return Backend.get_instance().get().from_array(np.float32(any_array))
        any_array = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
        np = <module 'numpy' from '/home/alobo/.local/lib/python3.10/site-packages/numpy/__init__.py'>

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_opencl_backend.py:44, in OpenCLBackend.from_array(self=<pyclesperanto_prototype._tier0._opencl_backend.OpenCLBackend object>, *args=(array([[ 40.,  32., ..., 200., 200.],
       [ 5...   [ 96.,  88., ...,  48.,  48.]], dtype=float32),), **kwargs={})
     43 def from_array(self, *args, **kwargs):
---> 44     return OCLArray.from_array(*args, **kwargs)
        args = (array([[ 40.,  32., ..., 200., 200.],
       [ 56.,  40., ..., 216., 216.],
       ...,
       [ 80.,  80., ...,  48.,  48.],
       [ 96.,  88., ...,  48.,  48.]], dtype=float32),)
        kwargs = {}

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_pycl.py:69, in OCLArray.from_array(cls=<class 'pyclesperanto_prototype._tier0._pycl.OCLArray'>, arr=array([[ 40.,  32., ..., 200., 200.],
       [ 5...   [ 96.,  88., ...,  48.,  48.]], dtype=float32), *args=(), **kwargs={})
     66 @classmethod
     67 def from_array(cls, arr, *args, **kwargs):
     68     assert_supported_ndarray_type(arr.dtype.type)
---> 69     queue = get_device().queue
     70     return OCLArray.to_device(queue, prepare(arr), *args, **kwargs)

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_device.py:45, in get_device()
     43 def get_device() -> Device:
     44     """Get the current device GPU class."""
---> 45     return _current_device._instance or select_device()
        _current_device._instance = None

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_device.py:75, in select_device(name=None, dev_type=None, score_key=None, device_index=-1)
     71 except:
     72     pass
---> 75 device = filter_devices(name, dev_type, score_key)[device_index]
        device_index = -1
        name = None
        dev_type = None
        score_key = None
     76 if name is not None and name not in device.name:
     77     warnings.warn(f"No OpenCL device found with {name} in their name. Using {device.name} instead.")

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_device.py:107, in filter_devices(name=None, dev_type=None, score_key=None)
     95 """Filter devices based on various options
     96 
     97 :param name: First device that contains ``name`` will be returned, defaults to None
   (...)
    104 :rtype: List[cl.Device]
    105 """
    106 devices = []
--> 107 for platform in cl.get_platforms():
        cl.get_platforms = <nanobind.nb_func object at 0x7db13b2db010>
        cl = <module 'pyopencl' from '/home/alobo/miniconda3/envs/napari25/lib/python3.10/site-packages/pyopencl/__init__.py'>
    108     for device in platform.get_devices():
    109         if name and name.lower() in device.name.lower():

LogicError: clGetPlatformIDs failed: PLATFORM_NOT_FOUND_KHR

[aloboa]

Following advice by Marcelo Zoccoloer, I run:
(napari25) alobo@pop-os:~/BioImage-Analysis-and-Data-Processing-Workshop-2025$ conda install -c conda-forge ocl-icd-system
(note environment is active),

and then run napari again. Same error:

---

LogicError                                Traceback (most recent call last)
File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/napari_accelerated_pixel_and_object_classification/_dock_widget.py:161, in ObjectSegmentation.__init__.<locals>.train_clicked(*arg=(False,), **kwargs={})
    157     return
    159 first_image_layer = self.get_selected_images()[0]
--> 161 self.train(
        self = <napari_accelerated_pixel_and_object_classification._dock_widget.ObjectSegmentation object at 0x78025b3f15a0>
        num_object_annotation_spinner = <PyQt5.QtWidgets.QSpinBox object at 0x7802c8a304c0>
        self.feature_selector = <napari_accelerated_pixel_and_object_classification._dock_widget.FeatureSelector object at 0x7802c8a332e0>
        num_max_depth_spinner = <PyQt5.QtWidgets.QSpinBox object at 0x780271b00550>
        num_trees_spinner = <PyQt5.QtWidgets.QSpinBox object at 0x780271b00670>
        filename_edit = FileEdit(mode='w', value=PosixPath('ObjectSegmenter.cl'))
        show_classifier_statistics_checkbox = <PyQt5.QtWidgets.QCheckBox object at 0x780271b011b0>
        first_image_layer = <Image layer 'blobs' at 0x78027d066a10>
    162     self.get_selected_images_data(),
    163     self.get_selected_annotation_data(),
    164     num_object_annotation_spinner.value(),
    165     self.feature_selector.getFeatures(),
    166     num_max_depth_spinner.value(),
    167     num_trees_spinner.value(),
    168     str(filename_edit.value.absolute()).replace("\\", "/").replace("//", "/"),
    169     show_classifier_statistics_checkbox.isChecked(),
    170     first_image_layer.scale
    171 )

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/napari_accelerated_pixel_and_object_classification/_dock_widget.py:288, in ObjectSegmentation.train(self=<napari_accelerated_pixel_and_object_classification._dock_widget.ObjectSegmentation object>, images=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8), annotation=array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, ...., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8), object_annotation_value=2, feature_definition='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', num_max_depth=2, num_trees=100, filename='/home/alobo/BioImage-Analysis-and-Data-Processing-Workshop-2025/ObjectSegmenter.cl', show_classifier_statistics=False, scale=array([1., 1.]))
    284     classifier.output_probability_of_class = object_annotation_value
    286 print("annotation shape", annotation.shape)
--> 288 classifier.train(feature_definition, annotation, images)
        classifier = Classifier type: ObjectSegmenter
--- Random forest info ---
Used features for training: gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1
Ground truth dimensions: 2
Maximum depth: 2
Number of ensembles: 100
Number of classes: 
Number of features: 0
Number of channels: 0Positive class identifier: 2
        feature_definition = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        annotation = array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, 0],
       ...,
       [0, 0, ..., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8)
        images = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
    290 print("Training done. Applying model...")
    292 result = np.asarray(classifier.predict(features=feature_definition, image=images))

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_object_segmenter.py:38, in ObjectSegmenter.train(self=Classifier type: ObjectSegmenter
--- Random fore...Number of channels: 0Positive class identifier: 2, features='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', ground_truth=array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, ...., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8), image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8), continue_training=False)
     31 """Train the classifier with.
     32 
     33 See Also
     34 --------
     35 .. PixelClassifier.train()
     36 """
     37 self.positive_class_identifier_from_file = self.positive_class_identifier
---> 38 super().train(features, ground_truth, image, continue_training=continue_training)
        features = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        ground_truth = array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, 0],
       ...,
       [0, 0, ..., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8)
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
        continue_training = False

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_pixel_classifier.py:101, in PixelClassifier.train(self=Classifier type: ObjectSegmenter
--- Random fore...Number of channels: 0Positive class identifier: 2, features='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', ground_truth=array([[0, 0, ..., 0, 0],
       [0, 0, ..., 0, ...., 0, 0],
       [0, 0, ..., 0, 0]], dtype=uint8), image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8), continue_training=False)
     99 # make features and convert in the right format
    100 self.num_ground_truth_dimensions = len(ground_truth.shape)
--> 101 features = self._make_features_potentially_multichannel(features, image)
        features = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        self = Classifier type: ObjectSegmenter
--- Random forest info ---
Used features for training: gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1
Ground truth dimensions: 2
Maximum depth: 2
Number of ensembles: 100
Number of classes: 
Number of features: 0
Number of channels: 0Positive class identifier: 2
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
    102 self.num_features = len(features)
    103 X, y = self._to_np(features, ground_truth)

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_pixel_classifier.py:282, in PixelClassifier._make_features_potentially_multichannel(self=Classifier type: ObjectSegmenter
--- Random fore...Number of channels: 0Positive class identifier: 2, features='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8))
    280     features = new_features
    281 else:
--> 282     features = self._make_features(features, image)
        features = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
        self = Classifier type: ObjectSegmenter
--- Random forest info ---
Used features for training: gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1
Ground truth dimensions: 2
Maximum depth: 2
Number of ensembles: 100
Number of classes: 
Number of features: 0
Number of channels: 0Positive class identifier: 2
    283 return features

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_pixel_classifier.py:304, in PixelClassifier._make_features(self=Classifier type: ObjectSegmenter
--- Random fore...Number of channels: 0Positive class identifier: 2, features='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1', image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8))
    302     if image is None:
    303         raise TypeError("If features are provided as string, an image must be given as well to generate features.")
--> 304     features = generate_feature_stack(image, features)
        features = 'gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1'
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
    306 return features

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/apoc/_utils.py:27, in generate_feature_stack(image=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8), features_specification='gaussian_blur=1 difference_of_gaussian=1 laplace_box_of_gaussian_blur=1 sobel_of_gaussian_blur=1')
      9 def generate_feature_stack(image, features_specification : Union[str, PredefinedFeatureSet] = None):
     10     """
     11     Creates a feature stack from a given image.
     12 
   (...)
     24     a list of OCLarray images
     25     """
---> 27     image = cle.push(image)
        image = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
        cle = <module 'pyclesperanto_prototype' from '/home/alobo/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/__init__.py'>
     29     # default features
     30     if features_specification is None:

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_push.py:41, in push(any_array=array([[ 40,  32, ..., 200, 200],
       [ 56,  ...,
       [ 96,  88, ...,  48,  48]], dtype=uint8))
     38 if hasattr(any_array, 'shape') and hasattr(any_array, 'dtype') and hasattr(any_array, 'get'):
     39     any_array = np.asarray(any_array.get())
---> 41 return Backend.get_instance().get().from_array(np.float32(any_array))
        any_array = array([[ 40,  32, ..., 200, 200],
       [ 56,  40, ..., 216, 216],
       ...,
       [ 80,  80, ...,  48,  48],
       [ 96,  88, ...,  48,  48]], dtype=uint8)
        np = <module 'numpy' from '/home/alobo/.local/lib/python3.10/site-packages/numpy/__init__.py'>

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_opencl_backend.py:44, in OpenCLBackend.from_array(self=<pyclesperanto_prototype._tier0._opencl_backend.OpenCLBackend object>, *args=(array([[ 40.,  32., ..., 200., 200.],
       [ 5...   [ 96.,  88., ...,  48.,  48.]], dtype=float32),), **kwargs={})
     43 def from_array(self, *args, **kwargs):
---> 44     return OCLArray.from_array(*args, **kwargs)
        args = (array([[ 40.,  32., ..., 200., 200.],
       [ 56.,  40., ..., 216., 216.],
       ...,
       [ 80.,  80., ...,  48.,  48.],
       [ 96.,  88., ...,  48.,  48.]], dtype=float32),)
        kwargs = {}

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_pycl.py:69, in OCLArray.from_array(cls=<class 'pyclesperanto_prototype._tier0._pycl.OCLArray'>, arr=array([[ 40.,  32., ..., 200., 200.],
       [ 5...   [ 96.,  88., ...,  48.,  48.]], dtype=float32), *args=(), **kwargs={})
     66 @classmethod
     67 def from_array(cls, arr, *args, **kwargs):
     68     assert_supported_ndarray_type(arr.dtype.type)
---> 69     queue = get_device().queue
     70     return OCLArray.to_device(queue, prepare(arr), *args, **kwargs)

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_device.py:45, in get_device()
     43 def get_device() -> Device:
     44     """Get the current device GPU class."""
---> 45     return _current_device._instance or select_device()
        _current_device._instance = None

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_device.py:75, in select_device(name=None, dev_type=None, score_key=None, device_index=-1)
     71 except:
     72     pass
---> 75 device = filter_devices(name, dev_type, score_key)[device_index]
        device_index = -1
        name = None
        dev_type = None
        score_key = None
     76 if name is not None and name not in device.name:
     77     warnings.warn(f"No OpenCL device found with {name} in their name. Using {device.name} instead.")

File ~/miniconda3/envs/napari25/lib/python3.10/site-packages/pyclesperanto_prototype/_tier0/_device.py:107, in filter_devices(name=None, dev_type=None, score_key=None)
     95 """Filter devices based on various options
     96 
     97 :param name: First device that contains ``name`` will be returned, defaults to None
   (...)
    104 :rtype: List[cl.Device]
    105 """
    106 devices = []
--> 107 for platform in cl.get_platforms():
        cl.get_platforms = <nanobind.nb_func object at 0x78027f137010>
        cl = <module 'pyopencl' from '/home/alobo/miniconda3/envs/napari25/lib/python3.10/site-packages/pyopencl/__init__.py'>
    108     for device in platform.get_devices():
    109         if name and name.lower() in device.name.lower():

LogicError: clGetPlatformIDs failed: PLATFORM_NOT_FOUND_KHR

[aloboa]

I have installed on another computer (Ubuntu 22.04.5 LTS).
The initial installation (following https://biapol.github.io/BioImage-Analysis-and-Data-Processing-Workshop-2025/intro.html#setting-up-the-environment) had the same issue, but here
$ conda install -c conda-forge ocl-icd-system
did the magic and I could get the blobs image processed, although with a warning:
/home/alobo2/.local/lib/python3.10/site-packages/numba/np/ufunc/parallel.py:371: NumbaWarning: The TBB threading layer requires TBB version 2021 update 6 or later i.e., TBB_INTERFACE_VERSION >= 12060. Found TBB_INTERFACE_VERSION = 12050. The TBB threading layer is disabled.!

This computer has a GPU. Is there a way to check within napari that the GPU is being used?

I will remove the environment and reinstall, with $ conda install -c conda-forge ocl-icd-system first, on the other computer and will post here.

[haesleinhuepf]

Hi @aloboa ,

in the menu Tools > Utilities > Select GPU (clesperanto) you can see which GPU is used and which GPUs are available.

If your GPU is not listed or causes trouble when using it, I recommend updating the GPU driver. You find links to the major GPU vendor driver websites here:
https://github.com/clEsperanto/pyclesperanto?tab=readme-ov-file#troubleshooting-graphics-cards-drivers

Let us know if this helps!

Best,
Robert

[aloboa]

I do not have Tools > Utilities > Select GPU (clesperanto) in my menu.

[haesleinhuepf]

I recommend updating the GPU driver. You find links to the major GPU vendor driver websites here:
https://github.com/clEsperanto/pyclesperanto?tab=readme-ov-file#troubleshooting-graphics-cards-drivers

[aloboa]

But my GPU works with cellpose and microsam.