Skip to content
Go to file

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

pix2pix for Map tiles

In this repository, we release various programs for image conversion using pix2pix-tensorflow for map tile images published on the web .

Operating environment

The operating environment is below.

Category version
OS Ubuntu 16.04 LTR 64bit
Framework Python 2.7.12
tensorflow 1.0.0
CUDA 8.0.61 (for GPU Environment)
CuDNN V5.1 (for GPU Environment)


  1. Install required packages
sudo apt-get update
sudo apt-get install python-pip python-dev libfreetype6-dev pkg-config liblapack-dev gfortran libtiff5-dev libjpeg8-dev zlib1g-dev liblcms2-dev libwebp-dev tcl8.6-dev tk8.6-dev python-tk python-scipy

sudo pip install \
    appdirs==1.4.0 \
    funcsigs==1.0.2 \
    google-api-python-client==1.6.2 \
    google-auth==0.7.0 \
    google-auth-httplib2==0.0.2 \
    google-cloud-core==0.22.1 \
    google-cloud-storage==0.22.0 \
    googleapis-common-protos==1.5.2 \
    httplib2==0.10.3 \
    mock==2.0.0 \
    numpy==1.12.0 \
    oauth2client==4.0.0 \
    packaging==16.8 \
    pbr==1.10.0 \
    protobuf==3.2.0 \
    pyasn1==0.2.2 \
    pyasn1-modules==0.0.8 \
    pyparsing==2.1.10 \
    rsa==3.4.2 \
    six==1.10.0 \
    uritemplate==3.0.0 \
  1. Install Tensorflow
  • No GPU Environment
sudo pip install --upgrade
  • With GPU Environment
sudo pip install --upgrade
  1. Coty and to

Then, you can use programes.

How to use

Set a Map tile fetching URL

First, You have to create JSON format file whicn include URL of fetching Map tile. A format of JSON file is below.

 "targetURL": {URL of target map tile}
 "inputURL": [
               {URL of training map tile 1}
               {URL of training map tile 2}

Parameter Discription
"targetURL" URL of target map tile
"inputURL" URL of training map tile. It is possible to set multipule URL in the "[ ]" divided by "," . 

Fromat of URL of fetching tile is below.

"url": base URL of fetching tile, 
 "type": type of map tile, 
 "format": format and parameter for fetching tile 
Parameter Discription
"url" Base URL of fetching tile
"type" Type of map tile. If you use WMST or "Map tile", set "tile". You can also set "wms" for OGS WMS Standard
"format" Setting for extention, tile address, parapeter for WMTS and WMS. Detail of setting, please refer setting samples in below.

Example for fetching tiled map format (GSI Japanese Aerial Photo (seamless))

"url": "", 
 "type": "tile", 
 "format": "{z}/{x}/{y}.jpg"
  • "format" for map tile type, plese set the tile coordinate oder (for example, {z}/{x}/{y}, {z}/{y}/{x}...), extention of tile, parameter of WMST and other. Please referr discripton of map tile provider.

Example for fetching WMS map tile (J-SHIS Land slide map WMS Service)

"url": "", 
"type": "wms", 
"format": "SERVICE=WMS&VERSION=1.3.0&REQUEST=GetMap&BBOX={minx},{miny},{maxx},{maxy}&CRS=EPSG:4612&WIDTH={output_width}&HEIGHT={output_height}&LAYERS=L-V3-S300&FORMAT=image/png&TRANSPARENT=FALSE"
  • "format" for WMS service, please set WMS request parameter. Please referr discripton of map tile provider.
  • BBOX, WIDHT and HEIGHT in WMS request, please set below.

Make training data set is used to create training data set. Parameters are below.

python "images_x_start" "images_x_end" "images_y_start" "images_y_end" "zoom_level"
                              --inputJson "INPUTJSON"
                              --outputPath "OUTPUTPATH"
Parameter Discription
"images_x_start" The address of the x direction of the tile which is the start of the fetching area
"images_x_end" The address of the x direction of the tile which is the end of the fetching area
"images_y_start" The address of the y direction of the tile which is the start of the fetching area
"images_y_end" The address of the x direction of the tile which is the end of the fetching area
--inputJson "INPUTJSON" Specify the json format file with the URL to get the map tile. The default is ". /jsonSample.txt"
--outputPath "OUTPUTPATH" Specify the output directory of the dataset. If there is no directory, it will be generated automatically. The default is "Data".
  • During execution, the original URL of the tile, the number of channels of input data (input channel), the number of channels of supervisor data (target channel), etc. are displayed. The number of each channel is input at the time of executing
  • After execution, the number of tiles in the directory specified by --outputPath will be generated by {sequential number}.tfrecords. Also, input_image{sequential number}.png and target_image{sequential number}.png are generated by connecting the acquired tiles in the directory where the program was executed.

Implementing the training

The training is performed using The execution format and main parameters are as follows  --input_dir "INPUT_DIR" 
                  --mode {train,test,export} 
                  --output_dir "OUTPUT_DIR"
                  --checkpoint "CHECKPOINT"
                  --max_steps "MAX_STEPS"
		  --max_epochs "MAX_EPOCHS"
                  --progress_freq "PROGRESS_FREQ"
		  --save_freq "SAVE_FREQ"
		  --ngf "NGF"
                  --input_ch "INPUT_CH"
		  --target_ch "TARGET_CH"
		  --GPUdevice "GPUDEVICE"		  
Parameter Discription
--input_dir 学習用データセットがあるディレクトリ。
--mode プログラムの実行モード。学習をする際には、"train"を指定。
--output_dir モデルの出力先ディレクトリ
--checkpoint 読み込む学習済みモデルがあるディレクトリ。指定しない場合、新規で学習を行なう。
--max_epochs 学習回数
--progress_freq 学習時に学習状況を表示するステップ数。デフォルトは50
--save_freq モデルを保存するステップ数。デフォルトは100。
--ngf 生成器の第一層のフィルター数。デフォルトは64。
--ndf 判別器の第一層のフィルター数。デフォルトは64。
--input_ch 入力データのチャンネル数。デフォルトは4。
--target_ch 教師データのチャンネル数。デフォルトは4。
--GPUdevice 実行するGPUの番号を指定します。デフォルトは0。複数のGPUがあるマシンを使用する場合に任意のGPUを指定して実行します。GPUがない場合はCPUを使用して実行されます。
  • 実行中に--save_freqで指定したステップごとに指定した保存先に学習モデル model-{ステップ数}.data -00000-of-00001、model-{ステップ数}.index、model-{ステップ数}.meta が出力されます。


This script is based on pix2pix-tensorflow. For more information about the program, see the distribution page of the underlying program.

We have also uploaded a sample study of the Ministry of the Environment's vegetation map to lad2veg using images from LANDSAT No. 7 and Geographical Survey map PNG elevation tiles. The results of --mode test can be viewed at


pix2pix for Map tiles




No releases published


No packages published
You can’t perform that action at this time.