AI - Media Art. 인공지능과 미디어아트

Media Art with AI
인공지능 기반 미디어아트 기술과 사례
2019.2
A.DAT - Open Media Art 전시 세미나
강태욱 공학박사
Ph.D Taewook, Kang
laputa99999@gmail.com
sites.google.com/site/bimprinciple

Media Art, Maker, Ph.D.
11 books author
TK. Kang

AI
A.DAT Open Media Art
Evolution of the interest to the Google research request “deep learning”. Obtained via
Google Trends (https://trends.google.com/trends/).

AI
CNN
(convolution neural network)
Deep Learning
Feature – classification
Learning

DL – deep learning
v = W·x + b

DL
y = φ(v)
머신러닝 딥러닝 신경망 개념, 종류 및 개발

DL - convolutional neural network

DL - convolutional neural network
ConvNetJS CIFAR-10

DL - Recurrent Neural Network & Long Short Term Memory
LSTM RNN Music Composition

DL - Generative Adversarial Network
Unsupervied Representation Learning
with Deep Convolutional Generative Adversarial Network

DL - Generative Adversarial Network
Image-to-Image Translation with Conditional Adversarial Networks

DL – Object Detection
딥러닝 기반 FAST 객체 탐색 기법 -
CNN, YOLO, SSD

DL
http://www.asimovinstitute.org/neural-
network-zoo/

DL from keras.models import Sequential
import keras
import numpy as np
from keras.applications import vgg16, inception_v3, resnet50, mobilenet
#VGG 모델을 로딩함
vgg_model = vgg16.VGG16(weights='imagenet')
from keras.preprocessing.image import load_img
from keras.preprocessing.image import img_to_array
from keras.applications.imagenet_utils import decode_predictions
import matplotlib.pyplot as plt
filename = '/home/ktw/tensorflow/door1.jpg'
# 이미지 로딩. PIL format
original = load_img(filename, target_size=(224, 224))
print('PIL image size',original.size)
plt.imshow(original)
plt.show()
# PIL 이미지를 numpy 배열로 변환
# Numpy 배열 (height, width, channel)
numpy_image = img_to_array(original)
plt.imshow(np.uint8(numpy_image))
plt.show()
print('numpy array size',numpy_image.shape)

DL
# 이미지를 배치 포맷으로 변환
# 데이터 학습을 위해 특정 축에 차원 추가
# 네트워크 형태는 batchsize, height, width, channels 이 됨
image_batch = np.expand_dims(numpy_image, axis=0)
print('image batch size', image_batch.shape)
plt.imshow(np.uint8(image_batch[0]))
# 모델 준비
processed_image = vgg16.preprocess_input(image_batch.copy())
# 각 클래스 속할 확률 예측
predictions = vgg_model.predict(processed_image)
# 예측된 확률을 클래스 라벨로 변환. 상위 5개 예측된 클래스 표시
label = decode_predictions(predictions)
print(label)

from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)
import tensorflow as tf
x = tf.placeholder(tf.float32, [None, 784])
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))
y = tf.nn.softmax(tf.matmul(x, W) + b)
y_ = tf.placeholder(tf.float32, [None, 10])
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y),
reduction_indices=[1]))
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)
sess = tf.InteractiveSession()
tf.global_variables_initializer().run()
for _ in range(1000):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_step, feed_dict={x: batch_xs, y_: batch_ys})
correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
print(sess.run(accuracy, feed_dict={x: mnist.test.images, y_:
mnist.test.labels}))
DL

DL
from keras.models import Sequential from keras.layers
import LSTM, Dense
import numpy as np
data_dim = 16
timesteps = 8
num_classes = 10 # expected input data shape: (batch_size, timesteps,
data_dim)
model = Sequential()
model.add(LSTM(32, return_sequences=True, input_shape=(timesteps,
data_dim))) # returns a sequence of vectors of dimension 32
model.add(LSTM(32, return_sequences=True)) # returns a sequence of
vectors of dimension 32 model.add(LSTM(32)) # return a single vector
of dimension 32
model.add(Dense(10, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='rmsprop',
metrics=['accuracy']) # Generate dummy training data x_train =
np.random.random((1000, timesteps, data_dim)) y_train =
np.random.random((1000, num_classes))
x_val = np.random.random((100, timesteps, data_dim)) y_val =
np.random.random((100, num_classes)) model.fit(x_train, y_train,
batch_size=64, epochs=5, validation_data=(x_val, y_val))

AI tools
AI
GPU
Open
data
Open
source
Collective
Intelligence
TPU

Open source
Richard Stallman
GNU's
Not Unix
‘84
OSI
SW
HW
Know
how
Data
Aca-
demy
Educa-
tion
NGO
Policy

Open source
http://guswnsxodlf.github.io/software-license
GNU General Public License(GPL) 2.0
– 의무 엄격. SW 수정 및 링크 경우 소스코드 제공 의무
GNU Lesser GPL(LGPL) 2.1
– 저작권 표시. LPGL 명시. 수정한 라이브러리 소스코드 공개
Berkeley Software Distribution(BSD) License
– 소스코드 공개의무 없음. 상용 SW 무제한 사용 가능
Apache License
– BSD와 유사. 소스코드 공개의무 없음
Mozilla Public License(MPL)
– 소스코드 공개의무 없음. 수정 코드는 MPL에 의해 배포
MIT License
– 라이선스 / 저작권만 명시 조건

AI Tool
Oxford Robotics Institute, 2017, Vote3Deep: Fast Object Detection in 3D
Point Clouds Using Efficient Convolutional Neural Networks, ICRA

AI Tool
Andrej Karpathy, 2015, The Unreasonable
Effectiveness of Recurrent Neural Networks

AI Tool
deepart.io

AI Tool
www.captionbot.ai

AI Tool
azure.microsoft.com/ko-kr/services/cognitive-
services/emotion

Open data - ImageNet
ImageNet Large Scale Visual Recognition Competition(ILSVRC)

Open data - ImageNet
케라스 기반 이미지 인식 딥러닝 모델 구현AlexNet, 2012
Top 5 test error
15.4%

Media Art + AI
Google Arts & CultureA.DAT Open Media Art

Media Art + AI
WDCH Dream, Refik Anadol StdudioA.DAT Open Media Art

Media Art + AI
Archive Dreaming, 2017
머신러닝 기반 170만 건 문서 처리
아카이브의 다차원 상호작용을 몰입형 설치 미디어로 표현
박물과 컨텍스트 관점에서 추억, 역사 문화를 재구성

Media Art + AI
Google’s Artists and Machine IntelligenceA.DAT Open Media Art

Media Art + AI
Deltu, 2016
'Deltu' uses two iPads to play mimicking games with a human opponent

Media Art + AI
Deep Learning Kubrick

Media Art + AI
Synthesizing Obama: Learning Lip Sync from Audio, 2017

Media Art + AI
Recognition, 2017

Media Art + AI
Portraits of Imaginary People

Media Art + AI
Kitty AI

Media Art + AI
Blade Runner—Autoencoded, 2016

Media Art + AI
flyAI, 2017

Media Art + AI
biometric mirror, and if you were perfect?
the artist Lucy McRae in her Biometric Mirror sits us in front of a mirror of a
futuristic beauty salon where the salon itself gives us a new image of ourselves,
born of an algorithm.

Media Art + AI
entangled, an infinite small space, 2018

Future of media art
ART
AI
MR
Robotics
IoT

강태욱, 2017, 머신러닝 딥러닝 신경망 개념, 종류 및 개발
강태욱, 2018, 케라스 기반 이미지 인식 딥러닝 모델 구현
ARS Electronica Festival 2017, Media Art between Natural and Artificial Intelligence, 2017
DAVID BROWEN, FLYAI, www.dwbowen.com/flyai
Lucy Mcrae, 2019.1, Biometric Mirror
ImageNet classification with Python and Keras
Keras Tutorial : Using pre-trained Imagenet models
Models for image classification with weights trained on ImageNet
Google, 텐서플로우 메뉴얼
YOLO: real-time object detection (paper)
ConvNetJS
carpedm20.github.io/faces
Reference

IoT – Embedded computer for prototyping

Cloud platform – MQTT, RaspberryPI, Blynk, ITFFF
Packing
Wireless
Sensor
Gateway
IoT
Control
Big data
analysis
Protocol
IoT
connection
service
A BIM ANALYSIS OF HVAC AND RADIANT COOLING SOLUTIONS, ROBERT CUBICK, 2016
KICT

AI - Media Art. 인공지능과 미디어아트

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to AI - Media Art. 인공지능과 미디어아트

Similar to AI - Media Art. 인공지능과 미디어아트 (20)

More from Tae wook kang

More from Tae wook kang (20)

Recently uploaded

Recently uploaded (20)

AI - Media Art. 인공지능과 미디어아트