Wolfram Neural Net Repository

Immediate Computable Access to Neural Net Models

CycleGAN Summer-to-Winter Translation

Turn a summertime photo into a wintertime photo

Released in 2017, this model exploits a novel technique for image translation, in which two models translating from A to B and vice versa are trained jointly with adversarial training. In addition to the adversarial loss, cycle consistency is also enforced in the loss function: when the output of the first translator is fed into the second, the final result is encouraged to match the input of the first translator. This allows successful training for image translation tasks in which only unpaired training data can be collected. This model was trained to translate summertime photos into wintertime photos.

Number of layers: 94 | Parameter count: 2,855,811 | Trained size: 12 MB |

Training Set Information

Examples

Resource retrieval

Get the pre-trained net:

In[1]:=
NetModel["CycleGAN Summer-to-Winter Translation"]
Out[1]=

Basic usage

Run the net on a photo:

In[2]:=
(* Evaluate this cell to get the example input *) CloudGet["https://www.wolframcloud.com/obj/db4dfe7c-5b5b-4dd4-a7ee-03beaedfc931"]
Out[2]=

Adapt to any size

Automatic image resizing can be avoided by replacing the net encoders. First get the net:

In[3]:=
net = NetModel["CycleGAN Summer-to-Winter Translation"]
Out[3]=

Get a photo:

In[4]:=
(* Evaluate this cell to get the example input *) CloudGet["https://www.wolframcloud.com/obj/2df00e08-022a-4db8-ac05-1b3d09567a97"]

Create a new encoder with the desired dimensions:

In[5]:=
netEnc = NetEncoder[{"Image", ImageDimensions[img]}]
Out[5]=

Attach the new net encoder and run the network:

In[6]:=
resizedNet = NetReplacePart[ net, {"Input" -> netEnc, "Output" -> NetDecoder[{"Image"}]}]
Out[6]=
In[7]:=
resizedNet[img]
Out[7]=

Net information

Inspect the sizes of all arrays in the net:

In[8]:=
NetInformation[ NetModel["CycleGAN Summer-to-Winter Translation"], "ArraysSizes"]
Out[8]=

Obtain the total number of parameters:

In[9]:=
NetInformation[ NetModel["CycleGAN Summer-to-Winter Translation"], \ "ArraysTotalElementCount"]
Out[9]=

Obtain the layer type counts:

In[10]:=
NetInformation[ NetModel["CycleGAN Summer-to-Winter Translation"], "LayerTypeCounts"]
Out[10]=

Display the summary graphic:

In[11]:=
NetInformation[ NetModel["CycleGAN Summer-to-Winter Translation"], "SummaryGraphic"]
Out[11]=

Export to MXNet

Export the net into a format that can be opened in MXNet:

In[12]:=
jsonPath = Export[FileNameJoin[{$TemporaryDirectory, "net.json"}], NetModel["CycleGAN Summer-to-Winter Translation"], "MXNet"]
Out[12]=

Export also creates a net.params file containing parameters:

In[13]:=
paramPath = FileNameJoin[{DirectoryName[jsonPath], "net.params"}]
Out[13]=

Get the size of the parameter file:

In[14]:=
FileByteCount[paramPath]
Out[14]=

The size is similar to the byte count of the resource object:

In[15]:=
ResourceObject["CycleGAN Summer-to-Winter Translation"]["ByteCount"]
Out[15]=

Represent the MXNet net as a graph:

In[16]:=
Import[jsonPath, {"MXNet", "NodeGraphPlot"}]
Out[16]=

Requirements

Wolfram Language 11.3 (March 2018) or above

Resource History

  • Date Created: 12 February 2018
  • Latest Update: 21 June 2018

Reference