visualize_concept_model

hybrid_learning.concepts.analysis.visualization.visualize_concept_model(handle, max_num_samples=10, label_templ=None, axes=None, start_idx=0, save_as=None, transform=None)[source]

Visualize predictions of a segmentation concept model on test samples. Visualization results will be shown using matplotlib.pyplot.imshow() functionality and are optionally saved to a file.

Parameters

  • handle (ConceptDetection2DTrainTestHandle) – the train/test handle for the concept model to evaluate (holds model and data)

  • start_idx (int) – relevant if axes is given; the row index to start filling axes

  • axes (Optional[Axes]) – optionally, one can specify an matplotlib.axes.Axes object to work on; generated by default

  • max_num_samples (int) – how many test samples to visualize

  • label_templ (Optional[str]) – template that is formatted and printed along with each sample; must contain a placeholder for the image ID as '{}'

  • save_as (Optional[str]) – .png file path to store the image file in using matplotlib.pyplot.savefig(); not saved if None

  • transform (Optional[Callable[[Any, Any], Tuple[Tensor, Tensor]]]) – a transformation to be applied to the tuple of model output and ground truth mask before mask visualization