keras topN显示,自编写代码案例

2023-07-22 03:23:03 447

对于使用已经训练好的模型，比如VGG,RESNET等，keras都自带了一个keras.applications.imagenet_utils.decode_predictions的方法,有很多限制：

defdecode_predictions(preds,top=5):
"""DecodesthepredictionofanImageNetmodel.

#Arguments
preds:Numpytensorencodingabatchofpredictions.
top:Integer,howmanytop-guessestoreturn.

#Returns
Alistoflistsoftopclasspredictiontuples
`(class_name,class_description,score)`.
Onelistoftuplespersampleinbatchinput.

#Raises
ValueError:Incaseofinvalidshapeofthe`pred`array
(mustbe2D).
"""
globalCLASS_INDEX
iflen(preds.shape)!=2orpreds.shape[1]!=1000:
raiseValueError('`decode_predictions`expects'
'abatchofpredictions'
'(i.e.a2Darrayofshape(samples,1000)).'
'Foundarraywithshape:'+str(preds.shape))
ifCLASS_INDEXisNone:
fpath=get_file('imagenet_class_index.json',
CLASS_INDEX_PATH,
cache_subdir='models',
file_hash='c2c37ea517e94d9795004a39431a14cb')
withopen(fpath)asf:
CLASS_INDEX=json.load(f)
results=[]
forpredinpreds:
top_indices=pred.argsort()[-top:][::-1]
result=[tuple(CLASS_INDEX[str(i)])+(pred[i],)foriintop_indices]
result.sort(key=lambdax:x[2],reverse=True)
results.append(result)
returnresults

把重要的东西挖出来，然后自己敲,这样就OK了，下例以MNIST数据集为例：

importkeras
fromkeras.modelsimportSequential
fromkeras.layersimportDense
importnumpyasnp
importtflearn
importtflearn.datasets.mnistasmnist

defdecode_predictions_custom(preds,top=5):
CLASS_CUSTOM=["0","1","2","3","4","5","6","7","8","9"]
results=[]
forpredinpreds:
top_indices=pred.argsort()[-top:][::-1]
result=[tuple(CLASS_CUSTOM[i])+(pred[i]*100,)foriintop_indices]
results.append(result)
returnresults

x_train,y_train,x_test,y_test=mnist.load_data(one_hot=True)

model=Sequential()
model.add(Dense(units=64,activation='relu',input_dim=784))
model.add(Dense(units=10,activation='softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='sgd',
metrics=['accuracy'])
model.fit(x_train,y_train,epochs=10,batch_size=128)
#score=model.evaluate(x_test,y_test,batch_size=128)
#print(score)
preds=model.predict(x_test[0:1,:])
p=decode_predictions_custom(preds)
for(i,(label,prob))inenumerate(p[0]):
print("{}.{}:{:.2f}%".format(i+1,label,prob))
#1.7:99.43%
#2.9:0.24%
#3.3:0.23%
#4.0:0.05%
#5.2:0.03%

补充知识：keras简单的去噪自编码器代码和各种类型自编码器代码

我就废话不多说了，大家还是直接看代码吧~

start=time()

fromkeras.modelsimportSequential
fromkeras.layersimportDense,Dropout,Input
fromkeras.layersimportEmbedding
fromkeras.layersimportConv1D,GlobalAveragePooling1D,MaxPooling1D
fromkerasimportlayers
fromkeras.modelsimportModel

#Parametersfordenoisingautoencoder
nb_visible=120
nb_hidden=64
batch_size=16
#Buildautoencodermodel
input_img=Input(shape=(nb_visible,))

encoded=Dense(nb_hidden,activation='relu')(input_img)
decoded=Dense(nb_visible,activation='sigmoid')(encoded)

autoencoder=Model(input=input_img,output=decoded)
autoencoder.compile(loss='mean_squared_error',optimizer='adam',metrics=['mae'])
autoencoder.summary()

#Train
###加一个early_stooping
importkeras

early_stopping=keras.callbacks.EarlyStopping(
monitor='val_loss',
min_delta=0.0001,
patience=5,
verbose=0,
mode='auto'
)
autoencoder.fit(X_train_np,y_train_np,nb_epoch=50,batch_size=batch_size,shuffle=True,
callbacks=[early_stopping],verbose=1,validation_data=(X_test_np,y_test_np))
#Evaluate
evaluation=autoencoder.evaluate(X_test_np,y_test_np,batch_size=batch_size,verbose=1)
print('val_loss:%.6f,val_mean_absolute_error:%.6f'%(evaluation[0],evaluation[1]))

end=time()
print('耗时：'+str((end-start)/60))

keras各种自编码代码

以上这篇kerastopN显示,自编写代码案例就是小编分享给大家的全部内容了，希望能给大家一个参考，也希望大家多多支持毛票票。

keras topN显示,自编写代码案例

热门推荐

随机推荐