Mirror

Pytorch CNN Visualisation Tool

This is a raw beta so expect lots of things to change and improve over time.

Getting started

To install mirror run

pip install git+https://github.com/FrancescoSaverioZuppichini/mirror.git

Basic example:

from mirror import mirror
from mirror.visualisations import *
from PIL import Image
from torchvision.models import resnet101, resnet18, vgg16, alexnet
from torchvision.transforms import ToTensor, Resize, Compose

# create a model
model = vgg16(pretrained=True)
# open some images
cat = Image.open("./cat.jpg")
dog_and_cat  = Image.open("./dog_and_cat.jpg")
# resize the image and make it a tensor
to_input = Compose([Resize((224,224)), ToTensor()])
# call mirror with the inputs and the model
mirror([to_input(cat), to_input(dog_and_cat)], model, visualisations=[DeepDreamVis, BackPropVis, GradCamVis]

It will automatic open a new tab in your browser

On the left you can see your model tree structure, by clicking on one layer all his children are showed. On the right there are the visualisation settings. You can select your input by clicking on the bottom tab.

Available Visualisations

Weights

Deep Dream

Back Prop / Guide Back Prop

By clicking on the radio button 'guide', all the relus negative output will be set to zero producing a nicer looking image

Grad Cam / Guide Grad Cam

Add text field for class

Create a Visualisation

You can find an example below

from mirror.visualisations.Visualisation import Visualisation

class DummyVisualisation(Visualisation):

    def __call__(self, inputs, layer):
        return inputs.repeat(self.params['repeat']['value'],1, 1, 1)

    @property
    def name(self):
        return 'dummy'

    def init_params(self):
        return {'repeat' : {
                 'type' : 'slider',
                 'min' : 1,
                 'max' : 100,
                 'value' : 3,
                 'step': 1,
                 'params': {}
                 }}

The __call__ function is called each time you click a layer or change a value in the options on the right.

The init_params function returns a dictionary of options that will be showed on the right drawer of the application. For now only slider , textfield and radio are supported. For example, the GradCamVis __init_params__ looks like:

class GradCamVis(Visualisation):
    ...
    def init_params(self):
        return {'guide': {'type': 'radio',
                          'value': False
                          },
                'class': {
                    'type': 'textfield',
                    'label': 'id',
                    'value': None
                          }
                }

Change the front-end

All the front-end is developed usin React and Material-UI, two very known frameworks, making easier for anybody to contribuite.

You can customise the front-end by changing the source code in mirror/client. After that, you need to build the react app and move the file to the server static folder.

I was not able to serve the static file directly from the /mirror/client/build folder if you know how to do it any pull request is welcome :)

cd ./mirror/mirror/client // assuming the root folder is called mirror
npm run build

Then you need to move the fiels from the mirror/mirror/client/build folder to mirror/mirror. You can remove all the files in mirror/mirro/static

mv ./build/static ../ && cp ./build/* ../static/

niezixu/mirror