keravis API Documentation
Convolutional layer activations
conv_layer_activations(model, layer, test_img, nested_model=None, title=None)Visualize activations of layer corresponding to test_img in a grid
- Args:
- modelkeras.Model
Model.
- layerstr
Layer whose activations to visualize.
- test_imgndarray
Image for which to look at activations of.
- nested_modelstr, default=None
Name of nested model, if any.
- titlestr, default=None
Title of the figure.
2-dimensional feature space representations
feature_space(model, dataset=None, X=None, y=None, kind='tsne', title=None)Visualize feature space of model on a set of images X in 2-dimensional space using tSNE or PCA
- Args:
- modelkeras.Model
Model.
- datasetkeras.preprocessing.image.DataIterator, default=None
Batched dataset. If given, X and y are ignored.
- Xndarray, default=None
Set of images.
- yndarray, default=None
Set of labels.
- kindstr, default=’tsne’
Type of plot. One of ‘tsne’ or ‘pca’.
- titlestr, default=None
Title of the figure.
Saliency maps
saliency_backprop(model, test_img, class_idx=0, title=None)Visualize the saliency map of test_img using vanilla backprop
- Args:
- modelkeras.Model
Model.
- test_imgndarray
Image for which to find saliency map of.
- class_idxint, default=0
Class index of image.
- titlestr, default=None
Title of the figure.
saliency_guided_backprop(model, test_img, class_idx=0, title=None)Visualize the saliency map of test_img using guided backprop
- Args:
- modelkeras.Model
Model.
- test_imgndarray
Image for which to find saliency map of.
- class_idxint, default=0
Class index of image.
- titlestr, default=None
Title of the figure.
saliency_occlusion(model, test_img, class_idx=0, title=None)Visualize the saliency map of test_img using occlusion
- Args:
- modelkeras.Model
Model.
- test_imgndarray
Image for which to find saliency map of.
- class_idxint, default=0
Class index of image.
- titlestr, default=None
Title of the figure.
Generated image that maximally activates classifier output
maximal_class_score_input(model, class_idx, dim, title=None)Visualize a generated image corresponding to a maximal class score of class_idx
- Args:
- modelkeras.Model
Model.
- class_idxint
Class index for which to find maximally activating image.
- dimtuple
(width,height,channels) of generated image.
- titlestr, default=None
Title of the figure.
Patches in a set of images that maximally activate an intermediate neuron
maximally_activating_patches(model, layer, dataset=None, X=None, nested_model = None, channel=None, title=None)Visualizes maximally activating patches in X of a random intermediate neuron in layer, channel
- Args:
- modelkeras.Model
Model.
- layerstr
Layer whose activations to visualize.
- datasetkeras.preprocessing.image.DataIterator, default=None
Batched dataset. If given, X and y are ignored.
- Xndarray, default=None
Set of images.
- nested_modelstr, default=None
Name of nested model, if any.
- channelint, default=None
Channel index. If not given, channel is randomly sampled.
- titlestr, default=None
Title of the figure.