map-GANerator
2D Map Generator - Generative Adversarial Network
Prerequisites
Dataset
I wrote this script to get random 512x512 map images from mapgen2 and collect the dataset with 6000 maps.
Run
- Get the dataset from here.
- Clone this repo and run train script to train from start with your desired hyperparameters for dataset and models.
You can also run this notebook on colab for faster results.
Results
Model
I create a Deep Convolutional Generative Adversarial Network (DCGAN) using Tensorflow with the help of Keras.
Configurations
Resolution: 64px
Epochs: 50
Batch Size: 32
Buffer Size: 6000
Seed Size: 100
Generator
Model: "sequential"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense (Dense) (None, 4096) 413696
_________________________________________________________________
reshape (Reshape) (None, 4, 4, 256) 0
_________________________________________________________________
up_sampling2d (UpSampling2D) (None, 8, 8, 256) 0
_________________________________________________________________
conv2d (Conv2D) (None, 8, 8, 256) 590080
_________________________________________________________________
batch_normalization (BatchNo (None, 8, 8, 256) 1024
_________________________________________________________________
activation (Activation) (None, 8, 8, 256) 0
_________________________________________________________________
up_sampling2d_1 (UpSampling2 (None, 16, 16, 256) 0
_________________________________________________________________
conv2d_1 (Conv2D) (None, 16, 16, 256) 590080
_________________________________________________________________
batch_normalization_1 (Batch (None, 16, 16, 256) 1024
_________________________________________________________________
activation_1 (Activation) (None, 16, 16, 256) 0
_________________________________________________________________
up_sampling2d_2 (UpSampling2 (None, 32, 32, 256) 0
_________________________________________________________________
conv2d_2 (Conv2D) (None, 32, 32, 128) 295040
_________________________________________________________________
batch_normalization_2 (Batch (None, 32, 32, 128) 512
_________________________________________________________________
activation_2 (Activation) (None, 32, 32, 128) 0
_________________________________________________________________
up_sampling2d_3 (UpSampling2 (None, 64, 64, 128) 0
_________________________________________________________________
conv2d_3 (Conv2D) (None, 64, 64, 128) 147584
_________________________________________________________________
batch_normalization_3 (Batch (None, 64, 64, 128) 512
_________________________________________________________________
activation_3 (Activation) (None, 64, 64, 128) 0
_________________________________________________________________
conv2d_4 (Conv2D) (None, 64, 64, 3) 3459
_________________________________________________________________
activation_4 (Activation) (None, 64, 64, 3) 0
=================================================================
Discriminator
Model: "sequential"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
conv2d_5 (Conv2D) (None, 32, 32, 32) 896
_________________________________________________________________
leaky_re_lu (LeakyReLU) (None, 32, 32, 32) 0
_________________________________________________________________
dropout (Dropout) (None, 32, 32, 32) 0
_________________________________________________________________
conv2d_6 (Conv2D) (None, 16, 16, 64) 18496
_________________________________________________________________
zero_padding2d (ZeroPadding2 (None, 17, 17, 64) 0
_________________________________________________________________
batch_normalization_4 (Batch (None, 17, 17, 64) 256
_________________________________________________________________
leaky_re_lu_1 (LeakyReLU) (None, 17, 17, 64) 0
_________________________________________________________________
dropout_1 (Dropout) (None, 17, 17, 64) 0
_________________________________________________________________
conv2d_7 (Conv2D) (None, 9, 9, 128) 73856
_________________________________________________________________
batch_normalization_5 (Batch (None, 9, 9, 128) 512
_________________________________________________________________
leaky_re_lu_2 (LeakyReLU) (None, 9, 9, 128) 0
_________________________________________________________________
dropout_2 (Dropout) (None, 9, 9, 128) 0
_________________________________________________________________
conv2d_8 (Conv2D) (None, 9, 9, 256) 295168
_________________________________________________________________
batch_normalization_6 (Batch (None, 9, 9, 256) 1024
_________________________________________________________________
leaky_re_lu_3 (LeakyReLU) (None, 9, 9, 256) 0
_________________________________________________________________
dropout_3 (Dropout) (None, 9, 9, 256) 0
_________________________________________________________________
conv2d_9 (Conv2D) (None, 9, 9, 512) 1180160
_________________________________________________________________
batch_normalization_7 (Batch (None, 9, 9, 512) 2048
_________________________________________________________________
leaky_re_lu_4 (LeakyReLU) (None, 9, 9, 512) 0
_________________________________________________________________
dropout_4 (Dropout) (None, 9, 9, 512) 0
_________________________________________________________________
flatten (Flatten) (None, 41472) 0
_________________________________________________________________
dense_1 (Dense) (None, 1) 41473
=================================================================