keras predict memory swap increase indefinitely

Question

I implemented a classification program using keras. I have a big set of images and I would like to predict each image using a for loop.

However, every time a new image is computed the swap memory increases. I tried to delete all variables inside of the predict function (and I'm sure that it is inside of this function that there is a problem) but the memory still increases.

for img in images:
    predict(img, model, categ_par, gl_par)

and the corresponding function:

def predict(image_path, model, categ_par, gl_par):   
    print("[INFO] loading and preprocessing image...")

    orig = cv2.imread(image_path)  

    image = load_img(image_path, target_size=(gl_par.img_width, gl_par.img_height))  
    image = img_to_array(image)  

    # important! otherwise the predictions will be '0'  
    image = image / 255  

    image = np.expand_dims(image, axis=0)

    # build the VGG16 network
    if(categ_par.method == 'VGG16'):
        model = applications.VGG16(include_top=False, weights='imagenet')  

    if(categ_par.method == 'InceptionV3'):
        model = applications.InceptionV3(include_top=False, weights='imagenet')  

    # get the bottleneck prediction from the pre-trained VGG16 model  
    bottleneck_prediction = model.predict(image)  

    # build top model  
    model = Sequential()  
    model.add(Flatten(input_shape=bottleneck_prediction.shape[1:]))  
    model.add(Dense(256, activation='relu'))  
    model.add(Dropout(0.5))  
    model.add(Dense(categ_par.n_class, activation='softmax'))  

    model.load_weights(categ_par.top_model_weights_path)  

    # use the bottleneck prediction on the top model to get the final classification  
    class_predicted = model.predict_classes(bottleneck_prediction) 
    probability_predicted = (model.predict_proba(bottleneck_prediction))

    classe = pd.DataFrame(list(zip(categ_par.class_indices.keys(), list(probability_predicted[0])))).\
    rename(columns = {0:'type', 1: 'prob'}).reset_index(drop=True)
    #print(classe)
    del model
    del bottleneck_prediction
    del image
    del orig
    del class_predicted
    del probability_predicted

    return classe.set_index(['type']).T

Answer 1

If you are using TensorFlow backend you will be building a model for each img in the for loop. TensorFlow just keeps appending graph onto graph etc. which means memory just rises. This is a well known occurrence and must be dealt with during hyperparameter optimization when you will be building many models, but also here.

from keras import backend as K

and put this at the end of predict():

K.clear_session()

Or you can just build one model and feed that as input to the predict function so you are not building a new one each time.