Keras Color Classification

이거 정확도가 0.79위로는 올라가지 않는데, 아마도 정상이라고 생각합니다.

사람이 색에 대해 느끼는 감각이 매우 주관적이니까요.

​

#!/usr/bin/env python

# coding: utf-8

​

# In[38]:

​

​

#0. 사용할 패키지 불러오기

# multi-class classification with Keras

import pandas

from keras.models import Sequential

from keras.layers import Dense

from keras.wrappers.scikit_learn import KerasClassifier

from keras.utils import np_utils

from sklearn.model_selection import cross_val_score

from sklearn.model_selection import KFold

from sklearn.preprocessing import LabelEncoder

from sklearn.pipeline import Pipeline

from tensorflow import keras

import tensorflow as tf

import seaborn as sns

import matplotlib as plt

​

​

# In[39]:

​

​

# load dataset

dataframe = pandas.read_csv("color_dataset.csv", header=None)

dataframe.head(3)

​

​

# In[ ]:

​

​

​

​

​

# In[40]:

​

​

rows, cols = dataframe.shape

train_dataset = dataframe.sample(frac=0.8,random_state=0)

test_dataset = dataframe.drop(train_dataset.index)

​

tempframe = dataframe.iloc[:, :cols-1]

train_stats = tempframe.describe()

print(train_stats)

train_stats = train_stats.transpose()

print(train_stats)

​

# label index

lidx = str(cols-1)

print(lidx)

​

​

# In[41]:

​

​

train_labels = train_dataset.iloc[:,cols-1] # pop(cols-1) # '3'

test_labels = test_dataset.iloc[:,cols-1] # '3'

all_labels = dataframe.iloc[:, cols-1]

print(train_labels.head(3))

print(test_labels.head(3))

print(all_labels.head(3))

train_dataset = train_dataset.drop([cols-1],axis=1)

print(train_dataset.head(3))

test_dataset = test_dataset.drop([cols-1], axis=1)

print(test_dataset.head(3))

​

​

# In[ ]:

​

​

​

​

​

# In[42]:

​

​

# encode class values as integers

encoder = LabelEncoder()

encoder.fit(all_labels)

encoded_Y = encoder.transform(train_labels)

# convert integers to dummy variables (i.e. one hot encoded)

dummy_y = np_utils.to_categorical(encoded_Y) # this perform one-hot encoding...

print(encoded_Y[:3])

print(dummy_y[:3])

​

# convert classes to class number, then make one-hot encoded matrix

enc_test = encoder.transform(test_labels)

test_y = np_utils.to_categorical(enc_test)

​

​

# In[59]:

​

​

input_count = cols - 1

class_count = len(encoder.classes_)

print(input_count)

print(class_count)

​

def norm(x):

return (x - train_stats['mean']) / train_stats['std']

normed_train_data = norm(train_dataset)

normed_test_data = norm(test_dataset)

​

​

# In[63]:

​

​

# define baseline model

def model_1():

# create model

model = Sequential()

model.add(Dense(8, input_dim=input_count, activation='relu'))

model.add(Dense(class_count, activation='softmax'))

# Compile model

model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

return model

def model_2():

# create model

model = Sequential()

model.add(Dense(32, input_dim=input_count, activation='relu'))

model.add(Dense(32, input_dim=input_count, activation='relu'))

model.add(Dense(class_count, activation='softmax'))

# Compile model

adam = tf.keras.optimizers.Adam(learning_rate=0.01) #, beta_1=0.9, beta_2=0.999, amsgrad=False)

model.compile(loss='categorical_crossentropy', optimizer=adam, metrics=['accuracy'])

#adam = tf.keras.optimizers.Adam(learning_rate=0.02)

#model.compile(loss='mse', optimizer=adam, metrics=['accuracy'])

return model

​

​

# In[64]:

​

​

model = model_2()

#model.fit(train_data, encoded_Y, epochs=100, validation_split = 0.2, verbose=0, callbacks=[PrintDot()])

history = model.fit(normed_train_data, dummy_y, epochs=5000, batch_size=20000, validation_split = 0.2)

​

​

# In[65]:

​

​

#6. 모델 평가하기

scores = model.evaluate(normed_test_data,test_y)

print(scores)

​

​

# In[68]:

​

​

fig, loss_ax=plt.pyplot.subplots()

loss_ax.plot(history.history['loss'], 'y',label='train_loss')

loss_ax.set_xlabel('epoch')

loss_ax.set_ylabel('loss')

loss_ax.legend(loc = 'upper left')

plt.pyplot.show()