notebook.community

Edit and run

Train sea lion classifier with a convnet





In [1]:



    


INPUT_DIR = '../../input/kaggle-sea-lion/02/'
OUTPUT_DIR = '../../output/kaggle-sea-lion/07/'
IMAGE_DIMS = (84,84,3)#class0
#IMAGE_DIMS = (84,84,3)#class1
#IMAGE_DIMS = (56,56,3)#class2
#IMAGE_DIMS = (42,42,3)#class3
#IMAGE_DIMS = (26,26,3)#class4

TRAIN_CLASS = 0
SAVE_WEIGHTS_FILE = OUTPUT_DIR + 'last-weights-medium1-class0.h5'

Y_CHANGE = [[1,2,3,4,5],[0]]
TRAIN_WEIGHT_RAW = (1,0.2,0.2,0.2,0.2,0.2)
TRAIN_WEIGHT = (1,1)
TEST_WEIGHT_RAW = (1,0.2,0.2,0.2,0.2,0.2)
TEST_WEIGHT = (1,1)

SAVE_MODEL_FILE = None
LOAD_WEIGHTS_FILE = SAVE_WEIGHTS_FILE
LOAD_MODEL_FILE = None
INPUT_DATASET_NAME = 'lion-patches-0px'

RECREATE_OUTPUT_DIR = True
RUN_TRAINING = True

TRAIN_EPOCHS = 10
INPUT_RANGE = 1
DEBUG = False

BATCH_SIZE=48



    











In [2]:



    


%matplotlib inline
import numpy as np
import pandas as pd
import h5py
import matplotlib.pyplot as plt
import sklearn
import os
import glob

import keras
from keras.preprocessing.image import ImageDataGenerator
from keras import models

from modules.logging import logger
import modules.utils as utils
from modules.utils import Timer
import modules.logging
import modules.cnn as cnn
import modules.lions as lions



    


















    






Using TensorFlow backend.

Prepare

Prepare output dir





In [3]:



    


utils.mkdirs(OUTPUT_DIR, dirs=['tf-logs','weights'], recreate=RECREATE_OUTPUT_DIR)
modules.logging.setup_file_logger(OUTPUT_DIR + 'out.log')
TF_LOGS_DIR = OUTPUT_DIR + 'tf-logs/'
WEIGHTS_DIR = OUTPUT_DIR + 'weights/'
input_dataset_path = INPUT_DIR + utils.dataset_name(INPUT_DATASET_NAME, IMAGE_DIMS)

logger.info('Output dirs created')



    


















    






2017-06-11 13:39:45,548 INFO Output dirs created

Prepare train, validate and test data flows





In [4]:



    


logger.info('Using dataset ' + input_dataset_path + ' as input')
h5file = h5py.File(input_dataset_path, 'r')

#used for image augmentation (creating new images for balancing)
image_augmentation_generator = ImageDataGenerator(
        featurewise_center=False,
        samplewise_center=False,
        featurewise_std_normalization=False,
        samplewise_std_normalization=False,
        zca_whitening=False,
        rotation_range=359,
        width_shift_range=0,
        height_shift_range=0,
        shear_range=0,
        horizontal_flip=True,
        vertical_flip=True)

#applied to all images during training
image_randomize_generator = ImageDataGenerator(
        featurewise_center=False,
        samplewise_center=False,
        featurewise_std_normalization=False,
        samplewise_std_normalization=False,
        zca_whitening=False,
        rotation_range=359,
        width_shift_range=0,
        height_shift_range=0,
        shear_range=0,
        horizontal_flip=True,
        vertical_flip=True)

logger.info('preparing train data')
train_batch_generator = utils.BatchGeneratorXYH5(h5file, start_ratio=0, end_ratio=INPUT_RANGE)
train_balance_generator = utils.ClassBalancerGeneratorXY(train_batch_generator,
                                                         image_augmentation=image_augmentation_generator,
                                                         output_weight=TRAIN_WEIGHT_RAW,
                                                         max_augmentation_ratio=1,
                                                         max_undersampling_ratio=1,
                                                         enforce_max_ratios=False,
                                                         batch_size=BATCH_SIZE,
                                                         start_ratio=0, end_ratio=0.7)
train_image_generator = utils.ImageAugmentationXYGenerator(train_balance_generator, image_randomize_generator)
train_generator = utils.ClassBalancerGeneratorXY(train_image_generator,
                                                         image_augmentation=image_augmentation_generator,
                                                         output_weight=TRAIN_WEIGHT,
                                                         max_augmentation_ratio=1,
                                                         max_undersampling_ratio=1,
                                                         enforce_max_ratios=False,
                                                         batch_size=BATCH_SIZE,
                                                         start_ratio=0, end_ratio=1,
                                                         change_y=Y_CHANGE)
#train_generator = utils.ChangeXYGenerator(train_image_generator, categorical_to_boolean)
logger.info('train size=' + str(train_generator.size) + ' batches=' + str(train_generator.nr_batches))


logger.info('preparing valid data')
valid_batch_generator = utils.BatchGeneratorXYH5(h5file, start_ratio=0, end_ratio=INPUT_RANGE)
valid_balance_generator = utils.ClassBalancerGeneratorXY(valid_batch_generator,
                                                         image_augmentation=image_augmentation_generator,
                                                         output_weight=TEST_WEIGHT_RAW,
                                                         max_augmentation_ratio=1,
                                                         max_undersampling_ratio=1,
                                                         enforce_max_ratios=False,
                                                         batch_size=BATCH_SIZE,
                                                         start_ratio=0.7, end_ratio=0.85)
valid_generator = utils.ClassBalancerGeneratorXY(valid_balance_generator,
                                                         image_augmentation=image_augmentation_generator,
                                                         output_weight=TEST_WEIGHT,
                                                         max_augmentation_ratio=1,
                                                         max_undersampling_ratio=1,
                                                         enforce_max_ratios=False,
                                                         batch_size=BATCH_SIZE,
                                                         start_ratio=0, end_ratio=1,
                                                         change_y=Y_CHANGE)
logger.info('valid size=' + str(valid_generator.size) + ' batches=' + str(valid_generator.nr_batches))



logger.info('preparing test data')
test_batch_generator = utils.BatchGeneratorXYH5(h5file, start_ratio=0, end_ratio=INPUT_RANGE)
test_balance_generator = utils.ClassBalancerGeneratorXY(test_batch_generator,
                                                         image_augmentation=image_augmentation_generator,
                                                         output_weight=TEST_WEIGHT_RAW,
                                                         max_augmentation_ratio=1,
                                                         max_undersampling_ratio=1,
                                                         enforce_max_ratios=False,
                                                         batch_size=BATCH_SIZE,
                                                         start_ratio=0.85, end_ratio=1)
test_generator = utils.ClassBalancerGeneratorXY(test_balance_generator,
                                                         image_augmentation=image_augmentation_generator,
                                                         output_weight=TEST_WEIGHT,
                                                         max_augmentation_ratio=1,
                                                         max_undersampling_ratio=1,
                                                         enforce_max_ratios=False,
                                                         batch_size=BATCH_SIZE,
                                                         start_ratio=0, end_ratio=1,
                                                         change_y=Y_CHANGE)
logger.info('test size=' + str(test_generator.size) + ' batches=' + str(test_generator.nr_batches))

#FIXME when using 1 on end ratio size and nr_batches gets negative (h5 batch generator, not balancer...)



    


















    






2017-06-11 13:39:45,589 INFO Using dataset ../../input/kaggle-sea-lion/02/lion-patches-0px-84-84.h5 as input
2017-06-11 13:39:45,604 INFO preparing train data
2017-06-11 13:39:45,606 INFO loading input data for class distribution analysis...
2017-06-11 13:39:45,607 INFO loading Y from raw dataset
2017-06-11 13:39:45,609 INFO > [started] generator dump...










    






1600/2263









    






2017-06-11 13:39:46,004 INFO > [done]    generator dump (395.147 ms)
2017-06-11 13:39:46,007 INFO raw sample class distribution
2017-06-11 13:39:46,008 INFO 0: 78
2017-06-11 13:39:46,010 INFO 1: 45
2017-06-11 13:39:46,011 INFO 2: 675
2017-06-11 13:39:46,013 INFO 3: 150
2017-06-11 13:39:46,014 INFO 4: 281
2017-06-11 13:39:46,015 INFO 5: 1034
2017-06-11 13:39:46,017 INFO overall output samples per class: 90
2017-06-11 13:39:46,018 INFO augmentation/undersampling ratio per class
2017-06-11 13:39:46,020 INFO SETUP FLOW 0 0.7
2017-06-11 13:39:46,021 INFO calculating source range according to start/end range of the desired output..
2017-06-11 13:39:46,023 INFO output distribution for this flow
2017-06-11 13:39:46,024 INFO 0: 62 (1.15)
2017-06-11 13:39:46,026 INFO 1: 12 (0.40)
2017-06-11 13:39:46,027 INFO 2: 12 (0.03)
2017-06-11 13:39:46,028 INFO 3: 12 (0.12)
2017-06-11 13:39:46,029 INFO 4: 12 (0.06)
2017-06-11 13:39:46,031 INFO 5: 12 (0.02)
2017-06-11 13:39:46,034 INFO source range: 0-1836 (1836)
2017-06-11 13:39:46,035 INFO output range: 0-125 (125)
2017-06-11 13:39:46,036 INFO loading input data for class distribution analysis...
2017-06-11 13:39:46,038 INFO loading Y from raw dataset
2017-06-11 13:39:46,039 INFO > [started] generator dump...
2017-06-11 13:39:46,040 INFO starting new flow...
2017-06-11 13:39:46,559 INFO > [done]    generator dump (519.571 ms)
2017-06-11 13:39:46,562 INFO raw sample class distribution
2017-06-11 13:39:46,563 INFO 0: 67
2017-06-11 13:39:46,564 INFO 1: 58
2017-06-11 13:39:46,566 INFO overall output samples per class: 116
2017-06-11 13:39:46,567 INFO augmentation/undersampling ratio per class
2017-06-11 13:39:46,569 INFO SETUP FLOW 0 1
2017-06-11 13:39:46,570 INFO calculating source range according to start/end range of the desired output..
2017-06-11 13:39:46,571 INFO output distribution for this flow
2017-06-11 13:39:46,574 INFO 0: 115 (1.73)
2017-06-11 13:39:46,575 INFO 1: 116 (2.00)
2017-06-11 13:39:46,576 INFO source range: 0-123 (123)
2017-06-11 13:39:46,577 INFO output range: 0-231 (231)
2017-06-11 13:39:46,579 INFO train size=231 batches=5
2017-06-11 13:39:46,580 INFO preparing valid data
2017-06-11 13:39:46,582 INFO loading input data for class distribution analysis...
2017-06-11 13:39:46,583 INFO loading Y from raw dataset
2017-06-11 13:39:46,585 INFO > [started] generator dump...










    






1600/2263









    






2017-06-11 13:39:46,918 INFO > [done]    generator dump (333.349 ms)
2017-06-11 13:39:46,921 INFO raw sample class distribution
2017-06-11 13:39:46,923 INFO 0: 78
2017-06-11 13:39:46,924 INFO 1: 45
2017-06-11 13:39:46,926 INFO 2: 675
2017-06-11 13:39:46,927 INFO 3: 150
2017-06-11 13:39:46,929 INFO 4: 281
2017-06-11 13:39:46,931 INFO 5: 1034
2017-06-11 13:39:46,932 INFO overall output samples per class: 90
2017-06-11 13:39:46,934 INFO augmentation/undersampling ratio per class
2017-06-11 13:39:46,935 INFO SETUP FLOW 0.7 0.85
2017-06-11 13:39:46,936 INFO calculating source range according to start/end range of the desired output..
2017-06-11 13:39:46,938 INFO output distribution for this flow
2017-06-11 13:39:46,939 INFO 0: 13 (1.15)
2017-06-11 13:39:46,941 INFO 1: 2 (0.40)
2017-06-11 13:39:46,942 INFO 2: 2 (0.03)
2017-06-11 13:39:46,943 INFO 3: 2 (0.12)
2017-06-11 13:39:46,945 INFO 4: 2 (0.06)
2017-06-11 13:39:46,946 INFO 5: 2 (0.02)
2017-06-11 13:39:46,949 INFO source range: 1588-1894 (306)
2017-06-11 13:39:46,951 INFO output range: 126-152 (26)
2017-06-11 13:39:46,952 INFO loading input data for class distribution analysis...
2017-06-11 13:39:46,954 INFO loading Y from raw dataset
2017-06-11 13:39:46,955 INFO > [started] generator dump...
2017-06-11 13:39:46,956 INFO starting new flow...
2017-06-11 13:39:47,056 INFO > [done]    generator dump (100.959 ms)
2017-06-11 13:39:47,059 INFO raw sample class distribution
2017-06-11 13:39:47,060 INFO 0: 15
2017-06-11 13:39:47,062 INFO 1: 11
2017-06-11 13:39:47,063 INFO overall output samples per class: 22
2017-06-11 13:39:47,065 INFO augmentation/undersampling ratio per class
2017-06-11 13:39:47,066 INFO SETUP FLOW 0 1
2017-06-11 13:39:47,068 INFO calculating source range according to start/end range of the desired output..
2017-06-11 13:39:47,069 INFO output distribution for this flow
2017-06-11 13:39:47,071 INFO 0: 22 (1.47)
2017-06-11 13:39:47,072 INFO 1: 22 (2.00)
2017-06-11 13:39:47,074 INFO source range: 0-25 (25)
2017-06-11 13:39:47,075 INFO output range: 0-44 (44)
2017-06-11 13:39:47,076 INFO SETUP FLOW 0 25
2017-06-11 13:39:47,078 INFO calculating source range according to start/end range of the desired output..
2017-06-11 13:39:47,079 INFO output distribution for this flow
2017-06-11 13:39:47,080 INFO 0: 2225 (1.15)
2017-06-11 13:39:47,081 INFO 1: 450 (0.40)
2017-06-11 13:39:47,082 INFO 2: 450 (0.03)
2017-06-11 13:39:47,083 INFO 3: 450 (0.12)
2017-06-11 13:39:47,085 INFO 4: 450 (0.06)
2017-06-11 13:39:47,086 INFO 5: 450 (0.02)
2017-06-11 13:39:47,089 INFO source range: 0-2262 (2262)
2017-06-11 13:39:47,091 INFO output range: 0-4475 (4475)
2017-06-11 13:39:47,092 INFO valid size=44 batches=1
2017-06-11 13:39:47,094 INFO preparing test data
2017-06-11 13:39:47,095 INFO loading input data for class distribution analysis...
2017-06-11 13:39:47,097 INFO loading Y from raw dataset
2017-06-11 13:39:47,098 INFO > [started] generator dump...










    






1600/2263









    






2017-06-11 13:39:47,423 INFO > [done]    generator dump (324.211 ms)
2017-06-11 13:39:47,426 INFO raw sample class distribution
2017-06-11 13:39:47,427 INFO 0: 78
2017-06-11 13:39:47,429 INFO 1: 45
2017-06-11 13:39:47,430 INFO 2: 675
2017-06-11 13:39:47,431 INFO 3: 150
2017-06-11 13:39:47,433 INFO 4: 281
2017-06-11 13:39:47,434 INFO 5: 1034
2017-06-11 13:39:47,436 INFO overall output samples per class: 90
2017-06-11 13:39:47,437 INFO augmentation/undersampling ratio per class
2017-06-11 13:39:47,438 INFO SETUP FLOW 0.85 1
2017-06-11 13:39:47,440 INFO calculating source range according to start/end range of the desired output..
2017-06-11 13:39:47,441 INFO output distribution for this flow
2017-06-11 13:39:47,442 INFO 0: 13 (1.15)
2017-06-11 13:39:47,444 INFO 1: 2 (0.40)
2017-06-11 13:39:47,445 INFO 2: 2 (0.03)
2017-06-11 13:39:47,446 INFO 3: 2 (0.12)
2017-06-11 13:39:47,448 INFO 4: 2 (0.06)
2017-06-11 13:39:47,449 INFO 5: 2 (0.02)
2017-06-11 13:39:47,452 INFO source range: 1869-2090 (221)
2017-06-11 13:39:47,453 INFO output range: 153-179 (26)
2017-06-11 13:39:47,455 INFO loading input data for class distribution analysis...
2017-06-11 13:39:47,457 INFO loading Y from raw dataset
2017-06-11 13:39:47,458 INFO > [started] generator dump...
2017-06-11 13:39:47,459 INFO starting new flow...
2017-06-11 13:39:47,516 INFO > [done]    generator dump (58.177 ms)
2017-06-11 13:39:47,520 INFO raw sample class distribution
2017-06-11 13:39:47,522 INFO 0: 16
2017-06-11 13:39:47,523 INFO 1: 10
2017-06-11 13:39:47,524 INFO overall output samples per class: 20
2017-06-11 13:39:47,526 INFO augmentation/undersampling ratio per class
2017-06-11 13:39:47,527 INFO SETUP FLOW 0 1
2017-06-11 13:39:47,528 INFO calculating source range according to start/end range of the desired output..
2017-06-11 13:39:47,529 INFO output distribution for this flow
2017-06-11 13:39:47,530 INFO 0: 20 (1.25)
2017-06-11 13:39:47,532 INFO 1: 20 (2.00)
2017-06-11 13:39:47,533 INFO source range: 0-25 (25)
2017-06-11 13:39:47,535 INFO output range: 0-40 (40)
2017-06-11 13:39:47,536 INFO SETUP FLOW 0 25
2017-06-11 13:39:47,537 INFO calculating source range according to start/end range of the desired output..
2017-06-11 13:39:47,538 INFO output distribution for this flow
2017-06-11 13:39:47,540 INFO 0: 2225 (1.15)
2017-06-11 13:39:47,541 INFO 1: 450 (0.40)
2017-06-11 13:39:47,543 INFO 2: 450 (0.03)
2017-06-11 13:39:47,544 INFO 3: 450 (0.12)
2017-06-11 13:39:47,547 INFO 4: 450 (0.06)
2017-06-11 13:39:47,549 INFO 5: 450 (0.02)
2017-06-11 13:39:47,552 INFO source range: 0-2262 (2262)
2017-06-11 13:39:47,553 INFO output range: 0-4475 (4475)
2017-06-11 13:39:47,555 INFO test size=40 batches=1

















In [5]:



    


#logger.info('INPUT DATASET DATA')
#dataset_path = INPUT_DIR + utils.dataset_name(INPUT_DATASET_NAME, IMAGE_DIMS)
#with h5py.File(input_dataset_path, 'r') as h5file:
#    logger.info('generator')
#    input_generator = utils.BatchGeneratorXYH5(h5file, start_ratio=0.001, end_ratio=0.0012, batch_size=64)
#    X, Y = utils.dump_xy_to_array(input_generator.flow(), input_generator.size, x=True, y=True)
#    utils.show_images(X, image_labels=utils.onehot_to_label(Y), group_by_label=False, cols=10, is_bgr=True, size=2)
#
#    logger.info('x ' + str(np.shape(X)))
#    logger.info('y ' + str(np.shape(Y)))
#    logger.info(str(utils.class_distribution(Y)))

if(DEBUG):
    logger.info('BALANCE GENERATOR DATA')
    #dataset_path = INPUT_DIR + utils.dataset_name(INPUT_DATASET_NAME, IMAGE_DIMS)
    X_train, Y_train = utils.dump_xy_to_array(train_generator.flow(), train_generator.size, x=False, y=True)
    logger.info('y ' + str(np.shape(Y_train)))
    #logger.info(str(utils.class_distribution(Y_train)))

    for xs,ys in train_generator.flow():
        utils.show_images(xs, image_labels=ys, cols=10, is_bgr=True, size=2)
        break

Prepare CNN model





In [6]:



    


logger.info('Load CNN model')
#model = lions.convnet_alexnet2_lion_keras(IMAGE_DIMS)

model = None
if(LOAD_MODEL_FILE!=None and os.path.isfile(LOAD_MODEL_FILE)):
    with open(LOAD_MODEL_FILE, 'r') as model_file:
        my = model_file.read()
        model = models.model_from_yaml(my)
        logger.info('loaded model from file ' + LOAD_MODEL_FILE)
else:
    model = lions.convnet_medium1_boolean(IMAGE_DIMS)
    logger.info('loaded model from function convnet_medium1_single')
    

if(LOAD_WEIGHTS_FILE!=None and os.path.isfile(LOAD_WEIGHTS_FILE)):
    model.load_weights(LOAD_WEIGHTS_FILE)
    logger.info('Loaded model weights from ' + LOAD_WEIGHTS_FILE)

logger.info('Model prepared')



    


















    






2017-06-11 13:39:47,595 INFO Load CNN model
/notebooks/datascience-snippets/kaggle-sea-lion/modules/lions.py:164: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(64, (3, 3), activation="relu", kernel_initializer="glorot_uniform", padding="same")`
  model.add(convolutional.Conv2D(64, (3, 3), activation='relu', padding='same', init='glorot_uniform'))
/notebooks/datascience-snippets/kaggle-sea-lion/modules/lions.py:166: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(128, (3, 3), activation="relu", kernel_initializer="glorot_uniform", padding="same")`
  model.add(convolutional.Conv2D(128, (3, 3), activation='relu', padding='same', init='glorot_uniform'))
/notebooks/datascience-snippets/kaggle-sea-lion/modules/lions.py:168: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(256, (3, 3), activation="relu", kernel_initializer="glorot_uniform", padding="same")`
  model.add(convolutional.Conv2D(256, (3, 3), activation='relu', padding='same', init='glorot_uniform'))
/notebooks/datascience-snippets/kaggle-sea-lion/modules/lions.py:173: UserWarning: Update your `Dense` call to the Keras 2 API: `Dense(1024, activation="relu", kernel_initializer="glorot_uniform")`
  model.add(core.Dense(1024, activation='relu', init='glorot_uniform'))
/notebooks/datascience-snippets/kaggle-sea-lion/modules/lions.py:175: UserWarning: Update your `Dense` call to the Keras 2 API: `Dense(1024, activation="relu", kernel_initializer="glorot_uniform")`
  model.add(core.Dense(1024, activation='relu', init='glorot_uniform'))
2017-06-11 13:39:47,756 INFO loaded model from function convnet_medium1_single
2017-06-11 13:39:47,761 INFO Model prepared

Train model





In [7]:



    


if(RUN_TRAINING):
    logger.info('Starting CNN training...')
    history = model.fit_generator(train_generator.flow(),
                  steps_per_epoch = train_generator.nr_batches,
                  nb_epoch = TRAIN_EPOCHS,
                  callbacks = cnn.get_callbacks_keras(model, WEIGHTS_DIR, TF_LOGS_DIR),
                  validation_data = valid_generator.flow(), 
                  validation_steps = valid_generator.nr_batches,
                  verbose = 1)

    if(SAVE_MODEL_FILE!=None):
        with open(SAVE_MODEL_FILE, 'w') as model_file:
            model_file.write(model.to_yaml())
            logger.info('Saved last model to ' + SAVE_MODEL_FILE)
    
    if(SAVE_WEIGHTS_FILE!=None):
        model.save_weights(SAVE_WEIGHTS_FILE)
        logger.info('Saved last weights to ' + SAVE_WEIGHTS_FILE)



    


















    






2017-06-11 13:39:47,772 INFO Starting CNN training...










    






INFO:tensorflow:Summary name conv2d_1/kernel:0 is illegal; using conv2d_1/kernel_0 instead.










    






2017-06-11 13:39:47,978 INFO Summary name conv2d_1/kernel:0 is illegal; using conv2d_1/kernel_0 instead.










    






INFO:tensorflow:Summary name conv2d_1/kernel:0 is illegal; using conv2d_1/kernel_0 instead.










    






2017-06-11 13:39:47,984 INFO Summary name conv2d_1/kernel:0 is illegal; using conv2d_1/kernel_0 instead.










    






INFO:tensorflow:Summary name conv2d_1/bias:0 is illegal; using conv2d_1/bias_0 instead.










    






2017-06-11 13:39:47,988 INFO Summary name conv2d_1/bias:0 is illegal; using conv2d_1/bias_0 instead.










    






INFO:tensorflow:Summary name conv2d_1/bias:0 is illegal; using conv2d_1/bias_0 instead.










    






2017-06-11 13:39:47,995 INFO Summary name conv2d_1/bias:0 is illegal; using conv2d_1/bias_0 instead.










    






INFO:tensorflow:Summary name conv2d_2/kernel:0 is illegal; using conv2d_2/kernel_0 instead.










    






2017-06-11 13:39:48,001 INFO Summary name conv2d_2/kernel:0 is illegal; using conv2d_2/kernel_0 instead.










    






INFO:tensorflow:Summary name conv2d_2/kernel:0 is illegal; using conv2d_2/kernel_0 instead.










    






2017-06-11 13:39:48,007 INFO Summary name conv2d_2/kernel:0 is illegal; using conv2d_2/kernel_0 instead.










    






INFO:tensorflow:Summary name conv2d_2/bias:0 is illegal; using conv2d_2/bias_0 instead.










    






2017-06-11 13:39:48,011 INFO Summary name conv2d_2/bias:0 is illegal; using conv2d_2/bias_0 instead.










    






INFO:tensorflow:Summary name conv2d_2/bias:0 is illegal; using conv2d_2/bias_0 instead.










    






2017-06-11 13:39:48,018 INFO Summary name conv2d_2/bias:0 is illegal; using conv2d_2/bias_0 instead.










    






INFO:tensorflow:Summary name conv2d_3/kernel:0 is illegal; using conv2d_3/kernel_0 instead.










    






2017-06-11 13:39:48,023 INFO Summary name conv2d_3/kernel:0 is illegal; using conv2d_3/kernel_0 instead.










    






INFO:tensorflow:Summary name conv2d_3/kernel:0 is illegal; using conv2d_3/kernel_0 instead.










    






2017-06-11 13:39:48,029 INFO Summary name conv2d_3/kernel:0 is illegal; using conv2d_3/kernel_0 instead.










    






INFO:tensorflow:Summary name conv2d_3/bias:0 is illegal; using conv2d_3/bias_0 instead.










    






2017-06-11 13:39:48,033 INFO Summary name conv2d_3/bias:0 is illegal; using conv2d_3/bias_0 instead.










    






INFO:tensorflow:Summary name conv2d_3/bias:0 is illegal; using conv2d_3/bias_0 instead.










    






2017-06-11 13:39:48,040 INFO Summary name conv2d_3/bias:0 is illegal; using conv2d_3/bias_0 instead.










    






INFO:tensorflow:Summary name dense_1/kernel:0 is illegal; using dense_1/kernel_0 instead.










    






2017-06-11 13:39:48,046 INFO Summary name dense_1/kernel:0 is illegal; using dense_1/kernel_0 instead.










    






INFO:tensorflow:Summary name dense_1/kernel:0 is illegal; using dense_1/kernel_0 instead.










    






2017-06-11 13:39:48,055 INFO Summary name dense_1/kernel:0 is illegal; using dense_1/kernel_0 instead.










    






INFO:tensorflow:Summary name dense_1/bias:0 is illegal; using dense_1/bias_0 instead.










    






2017-06-11 13:39:48,059 INFO Summary name dense_1/bias:0 is illegal; using dense_1/bias_0 instead.










    






INFO:tensorflow:Summary name dense_1/bias:0 is illegal; using dense_1/bias_0 instead.










    






2017-06-11 13:39:48,066 INFO Summary name dense_1/bias:0 is illegal; using dense_1/bias_0 instead.










    






INFO:tensorflow:Summary name dense_2/kernel:0 is illegal; using dense_2/kernel_0 instead.










    






2017-06-11 13:39:48,072 INFO Summary name dense_2/kernel:0 is illegal; using dense_2/kernel_0 instead.










    






INFO:tensorflow:Summary name dense_2/kernel:0 is illegal; using dense_2/kernel_0 instead.










    






2017-06-11 13:39:48,077 INFO Summary name dense_2/kernel:0 is illegal; using dense_2/kernel_0 instead.










    






INFO:tensorflow:Summary name dense_2/bias:0 is illegal; using dense_2/bias_0 instead.










    






2017-06-11 13:39:48,081 INFO Summary name dense_2/bias:0 is illegal; using dense_2/bias_0 instead.










    






INFO:tensorflow:Summary name dense_2/bias:0 is illegal; using dense_2/bias_0 instead.










    






2017-06-11 13:39:48,088 INFO Summary name dense_2/bias:0 is illegal; using dense_2/bias_0 instead.










    






INFO:tensorflow:Summary name dense_3/kernel:0 is illegal; using dense_3/kernel_0 instead.










    






2017-06-11 13:39:48,094 INFO Summary name dense_3/kernel:0 is illegal; using dense_3/kernel_0 instead.










    






INFO:tensorflow:Summary name dense_3/kernel:0 is illegal; using dense_3/kernel_0 instead.










    






2017-06-11 13:39:48,103 INFO Summary name dense_3/kernel:0 is illegal; using dense_3/kernel_0 instead.










    






INFO:tensorflow:Summary name dense_3/bias:0 is illegal; using dense_3/bias_0 instead.










    






2017-06-11 13:39:48,107 INFO Summary name dense_3/bias:0 is illegal; using dense_3/bias_0 instead.










    






INFO:tensorflow:Summary name dense_3/bias:0 is illegal; using dense_3/bias_0 instead.










    






2017-06-11 13:39:48,114 INFO Summary name dense_3/bias:0 is illegal; using dense_3/bias_0 instead.
/usr/local/lib/python3.4/dist-packages/ipykernel/__main__.py:9: UserWarning: Update your `fit_generator` call to the Keras 2 API: `fit_generator(<generator..., steps_per_epoch=5, validation_data=<generator..., verbose=1, epochs=10, callbacks=[<keras.ca..., validation_steps=1)`
2017-06-11 13:39:48,735 INFO starting new flow...
2017-06-11 13:39:48,738 INFO starting new flow...










    






Epoch 1/10
Epoch 1/10
4/5 [=======================>......] - ETA: 2s - loss: 1.1947 - acc: 0.5990 









    






2017-06-11 13:39:58,966 INFO starting new flow...
2017-06-11 13:39:58,968 INFO starting new flow...










    






Epoch 00000: val_acc improved from -inf to 0.25000, saving model to ../../output/kaggle-sea-lion/07/weights/weights-00-0.25.h5
5/5 [==============================] - 12s - loss: 1.1029 - acc: 0.5292 - val_loss: 0.6992 - val_acc: 0.2500
5/5 [==============================] - 12s - loss: 1.1029 - acc: 0.5292 - val_loss: 0.6992 - val_acc: 0.2500
Epoch 2/10
Epoch 2/10
4/5 [=======================>......] - ETA: 1s - loss: 0.6893 - acc: 0.5365Epoch 00001: val_acc improved from 0.25000 to 0.35417, saving model to ../../output/kaggle-sea-lion/07/weights/weights-01-0.35.h5
5/5 [==============================] - 11s - loss: 0.6761 - acc: 0.5958 - val_loss: 0.7925 - val_acc: 0.3542
5/5 [==============================] - 12s - loss: 0.6761 - acc: 0.5958 - val_loss: 0.7925 - val_acc: 0.3542
Epoch 3/10
Epoch 3/10
4/5 [=======================>......] - ETA: 1s - loss: 0.6275 - acc: 0.6562Epoch 00002: val_acc improved from 0.35417 to 0.75000, saving model to ../../output/kaggle-sea-lion/07/weights/weights-02-0.75.h5
5/5 [==============================] - 12s - loss: 0.6161 - acc: 0.6583 - val_loss: 0.5861 - val_acc: 0.7500
5/5 [==============================] - 12s - loss: 0.6161 - acc: 0.6583 - val_loss: 0.5861 - val_acc: 0.7500
Epoch 4/10
Epoch 4/10
4/5 [=======================>......] - ETA: 2s - loss: 0.6611 - acc: 0.6615Epoch 00003: val_acc did not improve
5/5 [==============================] - 11s - loss: 0.6723 - acc: 0.6250 - val_loss: 0.6176 - val_acc: 0.6458
5/5 [==============================] - 11s - loss: 0.6723 - acc: 0.6250 - val_loss: 0.6176 - val_acc: 0.6458
Epoch 5/10
Epoch 5/10
4/5 [=======================>......] - ETA: 2s - loss: 0.6367 - acc: 0.6250Epoch 00004: val_acc did not improve
5/5 [==============================] - 11s - loss: 0.6287 - acc: 0.6375 - val_loss: 0.7277 - val_acc: 0.4167
5/5 [==============================] - 11s - loss: 0.6287 - acc: 0.6375 - val_loss: 0.7277 - val_acc: 0.4167
Epoch 6/10
Epoch 6/10
4/5 [=======================>......] - ETA: 2s - loss: 0.6058 - acc: 0.6510Epoch 00005: val_acc did not improve
5/5 [==============================] - 11s - loss: 0.5948 - acc: 0.6708 - val_loss: 0.6680 - val_acc: 0.6042
5/5 [==============================] - 11s - loss: 0.5948 - acc: 0.6708 - val_loss: 0.6680 - val_acc: 0.6042
Epoch 7/10
Epoch 7/10
4/5 [=======================>......] - ETA: 2s - loss: 0.6184 - acc: 0.6146Epoch 00006: val_acc did not improve
5/5 [==============================] - 11s - loss: 0.6096 - acc: 0.6167 - val_loss: 0.6454 - val_acc: 0.5833
5/5 [==============================] - 11s - loss: 0.6096 - acc: 0.6167 - val_loss: 0.6454 - val_acc: 0.5833
Epoch 8/10
Epoch 8/10
4/5 [=======================>......] - ETA: 2s - loss: 0.5230 - acc: 0.6771Epoch 00007: val_acc did not improve
5/5 [==============================] - 11s - loss: 0.5276 - acc: 0.6625 - val_loss: 0.8927 - val_acc: 0.5833
5/5 [==============================] - 11s - loss: 0.5276 - acc: 0.6625 - val_loss: 0.8927 - val_acc: 0.5833
Epoch 9/10
Epoch 9/10
4/5 [=======================>......] - ETA: 2s - loss: 0.6056 - acc: 0.7396Epoch 00008: val_acc did not improve
5/5 [==============================] - 10s - loss: 0.5998 - acc: 0.7375 - val_loss: 0.6058 - val_acc: 0.6875
5/5 [==============================] - 10s - loss: 0.5998 - acc: 0.7375 - val_loss: 0.6058 - val_acc: 0.6875
Epoch 10/10
Epoch 10/10
4/5 [=======================>......] - ETA: 2s - loss: 0.5036 - acc: 0.7448Epoch 00009: val_acc did not improve
5/5 [==============================] - 11s - loss: 0.4638 - acc: 0.7625 - val_loss: 0.6294 - val_acc: 0.6875
5/5 [==============================] - 11s - loss: 0.4638 - acc: 0.7625 - val_loss: 0.6294 - val_acc: 0.6875










    






2017-06-11 13:41:45,547 INFO Saved last weights to ../../output/kaggle-sea-lion/07/last-weights-medium1-class0.h5

Epoch accuracy/loss





In [8]:



    


if(RUN_TRAINING):
    logger.info('Training info')
    cnn.show_training_info_keras(history)



    


















    






2017-06-11 13:41:45,559 INFO Training info

Confusion matrix





In [9]:



    


cnn.evaluate_dataset_keras(test_generator.flow(), 
                       test_generator.nr_batches, 
                       test_generator.size, 
                       model, 
                       class_labels=['any', 'male'])



    


















    






2017-06-11 13:41:45,812 INFO Evaluating model performance (40 samples)...
2017-06-11 13:41:45,821 INFO starting new flow...
2017-06-11 13:41:45,824 INFO starting new flow...
2017-06-11 13:41:46,875 INFO Accuracy: 0.854166686535 - Loss: 0.325673669577
2017-06-11 13:41:46,876 INFO Predicting Y for detailed analysis...
/notebooks/datascience-snippets/kaggle-sea-lion/modules/cnn.py:59: FutureWarning: comparison to `None` will result in an elementwise object comparison in the future.
  if(self.y_ds==None):
2017-06-11 13:41:48,362 INFO Accuracy: 0.5
2017-06-11 13:41:48,364 INFO Number of test samples: 40
2017-06-11 13:41:48,365 INFO Kappa score: 0.137931034483 (-1 bad; 0 just luck; 1 great)
2017-06-11 13:41:48,367 INFO 
             precision    recall  f1-score   support

        any       1.00      0.17      0.29        24
       male       0.44      1.00      0.62        16

avg / total       0.78      0.50      0.42        40

2017-06-11 13:41:48,368 INFO Accuracy per class:
2017-06-11 13:41:48,370 INFO any: 100.0%
2017-06-11 13:41:48,371 INFO male: 44.4%
2017-06-11 13:41:48,372 INFO Confusion matrix:
2017-06-11 13:41:48,374 INFO 
[[ 4 20]
 [ 0 16]]










    
























In [10]:



    


if(DEBUG):
    a = test_generator.flow()
    cnn.show_predictions(a, 50, model)



    











In [ ]:



    






    











In [ ]:



    






    











In [ ]:



    






    











In [ ]:



    






    











In [ ]:



    






    











In [ ]:

Content source: flaviostutz/datascience-snippets

Similar notebooks:

notebook.community | gallery | about