Why my Keras Image classification prediction always return one value...? - tensorflow

Question

I am using CNN to classify multiple people. People's number is 114 and each has about 120 image dataset.
I use keras and model that is used in dog vs cat classification.
It works well in Dogs and Cat classification.
But this is only return same value in different person image.
It says accuracy 99.12 but not work.
this is my train code
# coding: utf-8
# In[1]:
import os
import cv2
import numpy as np
from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import OneHotEncoder
from sklearn.cross_validation import train_test_split
from keras import backend as K
from keras import layers, models, optimizers
from keras.preprocessing.image import ImageDataGenerator, img_to_array, array_to_img
from keras.applications.resnet50 import preprocess_input
from keras.utils import np_utils
from numpy import array
# In[2]:
TRAIN_DIR = './Dataset'
train_folder_list = array(os.listdir(TRAIN_DIR))
# In[13]:
X = []
Y = []
# In[14]:
label_encoder = LabelEncoder()
integer_encoded = label_encoder.fit_transform(train_folder_list)
onehot_encoder = OneHotEncoder(sparse=False)
integer_encoded = integer_encoded.reshape(len(integer_encoded), 1)
onehot_encoded = onehot_encoder.fit_transform(integer_encoded)
for idx, index in enumerate(range(len(train_folder_list))):
path = os.path.join(TRAIN_DIR, train_folder_list[index])
path = path + '/'
img_list = os.listdir(path)
for img in img_list:
img_path = os.path.join(path, img)
img = cv2.imread(img_path)
if img is not None:
img = cv2.resize(img, (100, 100))
img = img_to_array(img)
X.append(img)
Y.append(idx)
X = np.array(X, dtype="float") / 255.0
# In[15]:
(X_train, X_test, Y_train, Y_test) = train_test_split(X,Y,test_size=0.2, random_state = 42)
Y_train = np_utils.to_categorical(Y_train, len(train_folder_list))
Y_test = np_utils.to_categorical(Y_test, len(train_folder_list))
print(X_train[0])
# In[17]:
nb_train_samples = len(X_train)
nb_valivation_samples = len(X_test)
batch_size = 32
# In[18]:
print(K.image_data_format())
# In[19]:
model = models.Sequential()
model.add(layers.Conv2D(32, (3,3), input_shape=(100, 100, 3)))
model.add(layers.Activation('relu'))
model.add(layers.MaxPooling2D(pool_size=(2, 2)))
model.add(layers.Conv2D(32, (3, 3)))
model.add(layers.Activation('relu'))
model.add(layers.MaxPooling2D(pool_size=(2, 2)))
model.add(layers.Conv2D(64, (3, 3)))
model.add(layers.Activation('relu'))
model.add(layers.MaxPooling2D(pool_size=(2, 2)))
model.add(layers.Flatten())
model.add(layers.Dense(500))
model.add(layers.Activation('relu'))
model.add(layers.Dropout(0.5))
model.add(layers.Dense(114))
model.add(layers.Activation('softmax'))
model.compile(loss='binary_crossentropy',
optimizer='rmsprop',
metrics=['accuracy'])
model.summary()
# In[20]:
train_datagen = ImageDataGenerator(
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
rescale=1. / 255,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='nearest')
test_datagen = ImageDataGenerator(
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
rescale=1. / 255,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='nearest')
# In[21]:
train_generator = train_datagen.flow(X_train, Y_train, batch_size=batch_size)
test_generator = test_datagen.flow(X_test , Y_test, batch_size=batch_size)
print(train_generator)
# In[ ]:
history = model.fit_generator(
train_generator,
steps_per_epoch=nb_train_samples // batch_size,
epochs=3,
validation_data=test_generator,
validation_steps=nb_valivation_samples // batch_size
)
model.save_weights('model_weights.h5')
model.save('model_keras.h5')
and my test code
from keras.models import load_model
import cv2
from keras.applications.resnet50 import preprocess_input
import numpy as np
model = load_model('model_keras.h5')
img = cv2.imread('input_test.jpg')
if img is not None:
img = cv2.resize(img, (100,100), interpolation=cv2.INTER_CUBIC)
img = np.reshape(img,[1,100,100,3])
img = preprocess_input(img)
classes = model.predict_classes(img)
print(classes)

Answer 1

Replace binary_crossentropy with categorical_crossentropy like this:
model.compile(loss='categorical_crossentropy',
optimizer='rmsprop',
metrics=['accuracy'])