softmax score by a cnn model

Question

I have build a CNN model for image classification. I want to pick five images that are correctly classified by the network and have the maximum softmax scores(for each class).

I have tried to check the model.evaluate(X_test,y_test) but it gives the overall softmax score of model.

model = Sequential()
model.add(Conv2D(96, (7, 7),  kernel_initializer='he_uniform', padding='same',strides=2, input_shape=(32, 32, 3)))
model.add(Activation("relu"))
model.add(Conv2D(64, (5, 5), kernel_initializer='he_uniform', padding='same',strides=2))
model.add(Activation("relu"))

model.add(Conv2D(128, (3, 3), kernel_initializer='he_uniform', padding='same',strides=2))
model.add(Activation("relu"))

model.add(MaxPooling2D(pool_size=(3, 3),padding='same',strides=3))

model.add(Flatten())

model.add(Dense(128))
model.add(Activation("relu"))

model.add(Dense(10, activation='softmax'))

compile model

opt = SGD(lr=0.0001, momentum=0.9)
model.compile(optimizer=opt, loss='categorical_crossentropy', metrics['accuracy'])

I want to see the softmax scores of correctly classified images.

If you have the overall softmax score and ground truth label, you can manually select the softmax scores of correctly classified ones. Does it make sense to you? — zihaozhihao
@zihaozhihao I have the overall sore of model, how can I get a score for single image — johntommond
You can get the output of last layer. Check this one github.com/keras-team/keras/issues/2495 — zihaozhihao

Dr. Snoopy Dr. Snoopy · Accepted Answer · 2019-09-28T01:17:56

0

votes

For this you have to use model.predict, not model.evaluate:

model.predict(X_test)

softmax score by a cnn model

compile model

1 Answers