Name

dnn_view - Visualize feature maps of stimuli.

Synopsis

dnn_view -net Net -layer Layer [Layer …] -chn Channel [Channel …] [-cmap Colormap] [-vmin Vmin] [-vmax Vmax] [-show] -out Output

Arguments

Required Arguments

Argument

Discription

net

Name of DNN Model, which should be placed in $DNNBRAIN_DATA/models with format *.pth.

layer

Convolution layers used to specify where activation is extracted from.E.g., ‘conv1’ represents the first convolution layerOnly support convolution layer!

chn

Channel numbers used to specify which channels are used to.Channel index starts from 1.Don’t support view more than 5 channels at once!

stim

A .stim.csv file which contains stimulus information

Optional Arguments

Argument

Discription

cmap

Show feature maps with the specified colormap.Default is ‘colorwarm’.

vmin

The minimal value used in colormap is applied in all feature maps.Default is the minimal value of each feature map for itself.

vmax

The maximal value used in colormap is applied in all feature maps.Default is the maximal value of each feature map for itself.

show

If used, display stimuli and feature maps in figures.

Outputs

A .jpg image visualizing the original stimulus and the feature maps of required net information.

Examples

dnn_view -net AlexNet -layer conv5 -chn 60 125 -stim ./test.stim.csv -show -out ./dnn_view.act.h5

The output image is displayed as below:

view