Currently I am trying to make a cnn that would allow for age detection on facial images. My dataset has the following shape where the images are grayscale.

(50000, 120, 120) - training 
(2983, 120, 120) - testing

And my model currently looks like the following - I've been testing/trying different methods.

    model = Sequential()
    model.add(Conv2D(64, kernel_size=3, use_bias=False,
                     input_shape=(size, size, 1)))

    model.add(Conv2D(32, kernel_size=3, use_bias=False))

    model.add(MaxPooling2D(pool_size=(2, 2)))

    model.add(Dense(128, use_bias=False))

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

    #TODO: Add in a lower learning rate - 0.001
    adam = optimizers.adam(lr=0.01)
    model.compile(optimizer=adam, loss='categorical_crossentropy',
    model.fit(x_train, y_train, validation_data=(x_test, y_test),
              epochs=number_of_epochs, verbose=1)

After running my data on just 10 epochs I started to initially see decent values but at the end of the run my results were the following and it has me concerned that my model is definitely over fitting.

How many epochs: 10
Train on 50000 samples, validate on 2939 samples
Epoch 1/10
50000/50000 [==============================] - 144s 3ms/step - loss: 1.7640 - acc: 0.3625 - val_loss: 1.6128 - val_acc: 0.4100
Epoch 2/10
50000/50000 [==============================] - 141s 3ms/step - loss: 1.5815 - acc: 0.4059 - val_loss: 1.5682 - val_acc: 0.4059
Epoch 3/10
50000/50000 [==============================] - 141s 3ms/step - loss: 1.5026 - acc: 0.4264 - val_loss: 1.6673 - val_acc: 0.4158
Epoch 4/10
50000/50000 [==============================] - 141s 3ms/step - loss: 1.3996 - acc: 0.4641 - val_loss: 1.5618 - val_acc: 0.4209
Epoch 5/10
50000/50000 [==============================] - 141s 3ms/step - loss: 1.2478 - acc: 0.5226 - val_loss: 1.6530 - val_acc: 0.4066
Epoch 6/10
50000/50000 [==============================] - 141s 3ms/step - loss: 1.0619 - acc: 0.5954 - val_loss: 1.6661 - val_acc: 0.4086
Epoch 7/10
50000/50000 [==============================] - 141s 3ms/step - loss: 0.8695 - acc: 0.6750 - val_loss: 1.7392 - val_acc: 0.3770
Epoch 8/10
50000/50000 [==============================] - 141s 3ms/step - loss: 0.7054 - acc: 0.7368 - val_loss: 1.8634 - val_acc: 0.3743
Epoch 9/10
50000/50000 [==============================] - 141s 3ms/step - loss: 0.5876 - acc: 0.7848 - val_loss: 1.8785 - val_acc: 0.3767
Epoch 10/10
50000/50000 [==============================] - 141s 3ms/step - loss: 0.5012 - acc: 0.8194 - val_loss: 2.2673 - val_acc: 0.3981
Model Saved

I assume the issue might be related to the number of images I have for each output class, but other then that I am a bit stuck in moving forward. Is there something wrong in my understanding/implementation? Any advice or critique would be well appreciated this is more of a learning project for me.


2 Answers 2


Try to use dropout after your dense layers not after maxpooling layers. Whatever comes before dense layers can be considered as the inputs of a classification layer. So keep them otherwise it somehow means you are loosing appropriate information. You should also be aware that you should not use dropout after the last layer.

Also you can add another dense layer, two hidden dense layers, for classification. It seems your data is not easy to learn.

  • $\begingroup$ Should I then move the dropout that's currently after maxpooling to be after the first dense layer? Or in general should I just put them after the last dense? And when you say not to use a dense layer after the last layer are you referring to the one with softmax? $\endgroup$ Dec 15, 2018 at 9:00
  • $\begingroup$ No, don't use them after pooling layers at all. In a CNN, You have convolutional stuff, then, you have dense layers. Suppose you have Dense layer #1, #2 and output. Use dropout after #1 and #2. $\endgroup$ Dec 15, 2018 at 9:04
  • $\begingroup$ And when you say not to use a dense layer after the last layer are you referring to the one with softmax? It was my mistake, I edited! $\endgroup$ Dec 15, 2018 at 9:07
  • 1
    $\begingroup$ I'm sorry I didnt see the edit, I will go and test out what you've mentioned hopefully it works out better. $\endgroup$ Dec 15, 2018 at 9:08

To deal with overfitting, you need to use regularization during the training:

  1. Weight regularization - The first thing you have to do (practically always) is to use regularization on the weights of the model. L1 or L2 regularization update the general loss function by adding another term known as the regularization term. As a result thee values of weights decrease because it assumes that a neural network with smaller weights leads to simpler models. Therefore, it will also reduce overfitting. If you are not sure what you need, just use L2.

    Keras - Usage of regularizers

  2. Dropout - Add dropout layers after dense layers (by the way, there are also advantages to using dropout after the convolution layers, it helps with occlusions). Just make sure not to use it at the final dense layer (the one with the same size as the number of classes).

  3. Data Augmentation - The simplest way to reduce overfitting is to increase the size of the training data. Use data augmentation to potentially expend your training set to "infinity". Keras's data augmentation is really simple an easy to use:

    Keras Image Preprocessing

If you implement these 3 steps, you will see drastic improvements (probably even just after the first one).

Further corrections and improvements (nothing to do with overfitting):

  • Your batch normalization layer should come after the non-linear activation, or more accurately, it needs to come before the next convolution layer.
  • Add an additional dense layer or 2 (only if the results are not good enough).
  • $\begingroup$ When you mention "Your batch normalization layer should come after the non-linear activation, or more accurately, it needs to come before the next convolution layer." isn't that what I'm already doing? As in adding it right after my first conv2d layer and before the second one? Or am I misunderstanding... $\endgroup$ Dec 16, 2018 at 0:33
  • $\begingroup$ also would the regularization go something like this in my context model.add(Dense(64, use_bias=False, kernel_regularizer=regularizers.l2( 0.01))) $\endgroup$ Dec 16, 2018 at 1:01
  • $\begingroup$ 1st comment: No, currently you are using the batch normalization before the non-linear activatoin: Conv->BatchNorm->ReLU. It needs to be Conv->ReLU->BatchNorm. $\endgroup$
    – Mark.F
    Dec 16, 2018 at 9:38
  • $\begingroup$ 2nd comment: Yes $\endgroup$
    – Mark.F
    Dec 16, 2018 at 9:38
  • $\begingroup$ Last comment, in Keras you can insert to ReLU activation as part of the CONV layer: model.add(Conv2D( 96, (11,11), padding='valid', kernel_regularizer=regularizers.l2(weight_decay), activation='relu')) $\endgroup$
    – Mark.F
    Dec 16, 2018 at 9:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.