def mask_blending_gan(offset_generator, mask_generator, discriminator, nb_fake=64, nb_real=32): assert len(mask_generator.input_shape) == 2 assert len(offset_generator.input_shape) == 2 g = Graph() mask_input_dim = mask_generator.input_shape[1] z_shape = (nb_fake, offset_generator.input_shape[1] - mask_input_dim) g.add_input(GAN.z_name, batch_input_shape=z_shape) g.add_node(Dense(mask_input_dim, activation='relu'), 'dense_mask', input=GAN.z_name) g.add_node(mask_generator, 'mask_generator', input='dense_mask') g.add_node(offset_generator, 'gen_offset', inputs=[GAN.z_name, 'dense_mask']) g.add_node(PyramidBlending(mask_generator, input_pyramid_layers=3, mask_pyramid_layers=2), GAN.generator_name, input='gen_offset') real_shape = (nb_real,) + g.nodes[GAN.generator_name].output_shape[1:] g.add_input(GAN.real_name, batch_input_shape=real_shape) g.add_node(discriminator, "discriminator", inputs=[GAN.generator_name, "real"], concat_axis=0) gan_outputs(g, fake_for_gen=(0, nb_fake), fake_for_dis=(nb_fake//2, nb_fake), real=(nb_fake, nb_fake+nb_real)) return g
def mask_blending_gan_new(offset_generator, mask_generator, discriminator, nb_fake=64, nb_real=32): assert len(mask_generator.input_shape) == 2 assert len(offset_generator.input_shape) == 2 g = Graph() mask_input_dim = mask_generator.input_shape[1] z_shape = (nb_fake, offset_generator.input_shape[1] - mask_input_dim) g.add_input(GAN.z_name, batch_input_shape=z_shape) g.add_node(Dense(32), 'gen_driver_dense_1', input=GAN.z_name) g.add_node(BatchNormalization(), 'gen_driver_bn_1', input='gen_driver_dense_1') g.add_node(Activation('relu'), 'gen_driver_act_1', input='gen_driver_bn_1') g.add_node(Dense(mask_input_dim), 'gen_driver_dense_2', input='gen_driver_act_1') g.add_node(BatchNormalization(), 'gen_driver_bn_2', input='gen_driver_dense_2') g.add_node(Layer(), 'driver', input='gen_driver_bn_2') # g.add_node(ZeroGradient(), 'gen_driver_zero_grad', input='driver') g.add_node(mask_generator, 'mask_generator', input='driver') g.add_node(offset_generator, 'gen_offset', input=GAN.z_name) g.add_node(PyramidBlending(mask_generator, input_pyramid_layers=3, mask_pyramid_layers=2), 'blending', input='gen_offset') reg_layer = Layer() act = ActivityInBoundsRegularizer(-1, 1) act.set_layer(reg_layer) reg_layer.regularizers = [act] g.add_node(reg_layer, GAN.generator_name, input='blending') real_shape = (nb_real, ) + g.nodes[GAN.generator_name].output_shape[1:] g.add_input(GAN.real_name, batch_input_shape=real_shape) g.add_node(discriminator, "discriminator", inputs=[GAN.generator_name, "real"], concat_axis=0) gan_outputs(g, fake_for_gen=(0, nb_fake - nb_real), fake_for_dis=(nb_fake - nb_real, nb_fake), real=(nb_fake, nb_fake + nb_real)) return g
def test_gan_get_config(tmpdir): z_shape = (1, 8, 8) z = Input(z_shape, name='z') g_out = Convolution2D(10, 2, 2, activation='relu', border_mode='same')(z) generator = Container(z, g_out) f, r = Input(z_shape, name='f'), Input(z_shape, name='r') dis_input = merge([f, r], mode='concat', concat_axis=1) dis_conv = Convolution2D(5, 2, 2, activation='relu')(dis_input) dis_flatten = Flatten()(dis_conv) dis = Dense(1, activation='sigmoid')(dis_flatten) discriminator = Container([f, r], gan_outputs(dis)) gan = GAN(generator, discriminator, z_shape, z_shape) weights_fname = str(tmpdir.mkdir("weights").join("{}.hdf5")) gan.save_weights(weights_fname) true_config = gan.get_config() import json with open(os.path.join(TEST_OUTPUT_DIR, "true_config.json"), 'w+') as f: json.dump(true_config, f, indent=2) gan_from_config = layer_from_config(true_config, custom_objects={ 'GAN': GAN, 'Split': Split, }) with open(os.path.join(TEST_OUTPUT_DIR, "loaded_config.json"), 'w+') as f: json.dump(gan_from_config.get_config(), f, indent=2) gan_from_config.load_weights(weights_fname)
def test_gan_custom_layer_graph(): z_shape = (1, 8, 8) z = Input(shape=z_shape, name='z') gen_cond = Input(shape=(1, 8, 8), name='gen_cond') inputs = [z, gen_cond] gen_input = merge(inputs, mode='concat', concat_axis=1) gen_output = Convolution2D(1, 2, 2, activation='relu', name='g1', border_mode='same')(gen_input) generator = Container(inputs, gen_output) f, r = Input(z_shape, name='fake'), Input(z_shape, name='real') inputs = [f, r] dis_input = merge(inputs, mode='concat', concat_axis=0) dis_conv = Convolution2D(5, 2, 2, name='d1', activation='relu')(dis_input) dis_flatten = Flatten()(dis_conv) dis = Dense(1, activation='sigmoid')(dis_flatten) discriminator = Container(inputs, gan_outputs(dis)) gan = GAN(generator, discriminator, z_shape=z_shape, real_shape=z_shape) gan.build('adam', 'adam', gan_binary_crossentropy) fn = gan.compile_custom_layers(['g1', 'd1']) z = np.random.uniform(-1, 1, (64,) + z_shape) real = np.random.uniform(-1, 1, (64,) + z_shape) cond = np.random.uniform(-1, 1, (64,) + z_shape) print(z.shape) print(real.shape) print(cond.shape) fn({'z': z, 'gen_cond': cond, 'real': real})
def get_discriminator(): f, r = Input(z_shape, name='f'), Input(z_shape, name='r') inputs = [f, r] dis_input = merge(inputs, mode='concat', concat_axis=1) dis_conv = Convolution2D(5, 2, 2, activation='relu')(dis_input) dis_flatten = Flatten()(dis_conv) dis = Dense(1, activation='sigmoid')(dis_flatten) return Container(inputs, gan_outputs(dis))
def discriminator(x): return gan_outputs(sequential([ Flatten(), Dense(1), ])(concat(x)), fake_for_gen=(0, 10), fake_for_dis=(0, 10), real=(10, 20))
def mask_blending_gan_new(offset_generator, mask_generator, discriminator, nb_fake=64, nb_real=32): assert len(mask_generator.input_shape) == 2 assert len(offset_generator.input_shape) == 2 g = Graph() mask_input_dim = mask_generator.input_shape[1] z_shape = (nb_fake, offset_generator.input_shape[1] - mask_input_dim) g.add_input(GAN.z_name, batch_input_shape=z_shape) g.add_node(Dense(32), 'gen_driver_dense_1', input=GAN.z_name) g.add_node(BatchNormalization(), 'gen_driver_bn_1', input='gen_driver_dense_1') g.add_node(Activation('relu'), 'gen_driver_act_1', input='gen_driver_bn_1') g.add_node(Dense(mask_input_dim), 'gen_driver_dense_2', input='gen_driver_act_1') g.add_node(BatchNormalization(), 'gen_driver_bn_2', input='gen_driver_dense_2') g.add_node(Layer(), 'driver', input='gen_driver_bn_2') # g.add_node(ZeroGradient(), 'gen_driver_zero_grad', input='driver') g.add_node(mask_generator, 'mask_generator', input='driver') g.add_node(offset_generator, 'gen_offset', input=GAN.z_name) g.add_node(PyramidBlending(mask_generator, input_pyramid_layers=3, mask_pyramid_layers=2), 'blending', input='gen_offset') reg_layer = Layer() act = ActivityInBoundsRegularizer(-1, 1) act.set_layer(reg_layer) reg_layer.regularizers = [act] g.add_node(reg_layer, GAN.generator_name, input='blending') real_shape = (nb_real,) + g.nodes[GAN.generator_name].output_shape[1:] g.add_input(GAN.real_name, batch_input_shape=real_shape) g.add_node(discriminator, "discriminator", inputs=[GAN.generator_name, "real"], concat_axis=0) gan_outputs(g, fake_for_gen=(0, nb_fake - nb_real), fake_for_dis=(nb_fake - nb_real, nb_fake), real=(nb_fake, nb_fake+nb_real)) return g
def mask_blending_gan(offset_generator, mask_generator, discriminator, nb_fake=64, nb_real=32): assert len(mask_generator.input_shape) == 2 assert len(offset_generator.input_shape) == 2 g = Graph() mask_input_dim = mask_generator.input_shape[1] z_shape = (nb_fake, offset_generator.input_shape[1] - mask_input_dim) g.add_input(GAN.z_name, batch_input_shape=z_shape) g.add_node(Dense(mask_input_dim, activation='relu'), 'dense_mask', input=GAN.z_name) g.add_node(mask_generator, 'mask_generator', input='dense_mask') g.add_node(offset_generator, 'gen_offset', inputs=[GAN.z_name, 'dense_mask']) g.add_node(PyramidBlending(mask_generator, input_pyramid_layers=3, mask_pyramid_layers=2), GAN.generator_name, input='gen_offset') real_shape = (nb_real, ) + g.nodes[GAN.generator_name].output_shape[1:] g.add_input(GAN.real_name, batch_input_shape=real_shape) g.add_node(discriminator, "discriminator", inputs=[GAN.generator_name, "real"], concat_axis=0) gan_outputs(g, fake_for_gen=(0, nb_fake), fake_for_dis=(nb_fake // 2, nb_fake), real=(nb_fake, nb_fake + nb_real)) return g
def simple_gan(): z = Input(batch_shape=simple_gan_z_shape, name='z') generator = sequential([ Dense(simple_gan_nb_z, activation='relu', name='g1'), Dense(simple_gan_nb_z, activation='relu', name='g2'), Dense(simple_gan_nb_out, activation='sigmoid', name='g3'), ])(z) fake = Input(batch_shape=simple_gan_real_shape, name='fake') real = Input(batch_shape=simple_gan_real_shape, name='real') discriminator = sequential([ Dense(20, activation='relu', input_dim=2, name='d1'), Dense(1, activation='sigmoid', name='d2') ])(concat([fake, real], axis=0)) return GAN(Container(z, generator), Container([fake, real], gan_outputs(discriminator)), simple_gan_z_shape[1:], simple_gan_real_shape[1:])
def test_gan_graph(): z_shape = (1, 8, 8) z = Input(shape=z_shape, name='z') gen_cond = Input(shape=(1, 8, 8), name='gen_cond') inputs = [z, gen_cond] gen_input = merge(inputs, mode='concat', concat_axis=1) gen_output = Convolution2D(10, 2, 2, activation='relu', border_mode='same')(gen_input) generator = Container(inputs, gen_output) f, r = Input(z_shape, name='f'), Input(z_shape, name='r') inputs = [f, r] dis_input = merge(inputs, mode='concat', concat_axis=1) dis_conv = Convolution2D(5, 2, 2, activation='relu')(dis_input) dis_flatten = Flatten()(dis_conv) dis = Dense(1, activation='sigmoid')(dis_flatten) discriminator = Container(inputs, gan_outputs(dis)) gan = GAN(generator, discriminator, z_shape=z_shape, real_shape=z_shape) gan.build('adam', 'adam', gan_binary_crossentropy) gan.compile() gan.generate({'gen_cond': np.zeros((64,) + z_shape)}, nb_samples=64)
def discriminator(x): return gan_outputs(sequential([ Flatten(), Dense(1), ])(concat(x)), fake_for_gen=(0, 10), fake_for_dis=(0, 10), real=(10, 20))