Backends (pymia.data.backends
package)¶
PyTorch¶
- class pymia.data.backends.pytorch.PytorchDatasetAdapter(*args: Any, **kwargs: Any)[source]¶
A wrapper class for
PymiaDatasource
to fit the torch.utils.data.Dataset interface.- Parameters
datasource (.PymiaDatasource) – The pymia datasource instance.
- class pymia.data.backends.pytorch.SubsetSequentialSampler(*args: Any, **kwargs: Any)[source]¶
Samples elements sequential from a given list of indices, without replacement.
The class adopts the torch.utils.data.Sampler interface.
- Parameters
list (indices) – list of indices that define the subset to be used for the sampling.
TensorFlow¶
- pymia.data.backends.tensorflow.get_tf_generator(data_source: pymia.data.extraction.datasource.PymiaDatasource)[source]¶
Returns a generator that wraps
PymiaDatasource
for the TensorFlow data handling.The returned generator can be used with tf.data.Dataset.from_generator in order to build a TensorFlow dataset.
- Parameters
data_source (.PymiaDatasource) – the datasource to be wrapped.
- Returns
Function that loops over the entire datasource and yields all entries.
- Return type
generator