def train():
from keras.metrics import categorical_accuracy
TransformImage.register('flatten', lambda img: img.flatten())
transform = (TransformImage(0)
.by('rerange', 0, 255, 0, 1, 'float32')
.by('flatten'))
build_batch = (BuildBatch(BATCH_SIZE)
.by(0, 'vector', 'float32')
.by(1, 'one_hot', 'uint8', NUM_CLASSES))
plot = PlotLines((0, 1), layout=(2, 1), every_sec=1)
print('loading data...')
train_samples, test_samples = load_samples()
print('creating network ...')
network = create_network()
print('training...', NUM_EPOCHS)
for epoch in range(NUM_EPOCHS):
print('EPOCH:', epoch)
t_loss, t_acc = (train_samples >> PrintProgress(train_samples) >>
transform >> build_batch >>
network.train() >> plot >> Unzip())
print('train loss : {:.6f}'.format(t_loss >> Mean()))
print('train acc : {:.1f}'.format(100 * (t_acc >> Mean())))
e_acc = (test_samples >> transform >> build_batch >>
network.evaluate([categorical_accuracy]))
print('test acc : {:.1f}'.format(100 * e_acc))
network.save_best(e_acc, isloss=False)
def train():
from keras.metrics import categorical_accuracy
rerange = TransformImage(0).by('rerange', 0, 255, 0, 1, 'float32')
build_batch = (BuildBatch(BATCH_SIZE).by(0, 'image', 'float32').by(
1, 'one_hot', 'uint8', NUM_CLASSES))
p = 0.1
augment = (AugmentImage(0).by('identical', 1.0).by(
'elastic', p, [5, 5], [100, 100],
[0, 100]).by('brightness', p,
[0.7, 1.3]).by('color', p, [0.7, 1.3]).by(
'shear', p,
[0, 0.1]).by('fliplr', p).by('rotate', p, [-10, 10]))
plot_eval = PlotLines((0, 1), layout=(2, 1))
print('creating network...')
network = create_network()
print('loading data...')
train_samples, test_samples = load_samples()
train_samples, val_samples = train_samples >> SplitRandom(0.8)
print('training...', len(train_samples), len(val_samples))
for epoch in range(NUM_EPOCHS):
print('EPOCH:', epoch)
t_loss, t_acc = (train_samples >> PrintProgress(train_samples) >>
Pick(PICK) >> augment >> rerange >> Shuffle(100) >>
build_batch >> network.train() >> Unzip())
t_loss, t_acc = t_loss >> Mean(), t_acc >> Mean()
print("train loss : {:.6f}".format(t_loss))
print("train acc : {:.1f}".format(100 * t_acc))
v_loss, v_acc = (val_samples >> rerange >> build_batch >>
network.validate() >> Unzip())
v_loss, v_acc = v_acc >> Mean(), v_acc >> Mean()
print('val loss : {:.6f}'.format(v_loss))
print('val acc : {:.1f}'.format(100 * v_acc))
network.save_best(v_acc, isloss=False)
plot_eval((t_acc, v_acc))
print('testing...', len(test_samples))
e_acc = (test_samples >> rerange >> build_batch >> network.evaluate(
[categorical_accuracy]))
print('test acc : {:.1f}'.format(100 * e_acc))
def train():
from keras.metrics import categorical_accuracy
print('creating network ...')
network = create_network()
print('loading data...')
train_samples, test_samples = load_samples()
augment = (AugmentImage(0)
.by('identical', 1)
.by('translate', 0.5, [-3, +3], [-3, +3])
.by('rotate', 0.5, [-5, +5])
.by('shear', 0.5, [0, 0.2])
.by('elastic', 0.5, [5, 5], [100, 100], [0, 100]))
transform = (TransformImage(0)
.by('rerange', 0, 255, 0, 1, 'float32'))
build_batch = (BuildBatch(BATCH_SIZE, prefetch=0)
.input(0, 'image', 'float32')
.output(1, 'one_hot', 'uint8', NUM_CLASSES))
plot = PlotLines((0, 1), layout=(2, 1), every_sec=1)
print('training...', NUM_EPOCHS)
for epoch in range(NUM_EPOCHS):
print('EPOCH:', epoch)
t_loss, t_acc = (train_samples >> PrintProgress(train_samples) >>
Pick(PICK) >> augment >> transform >> Shuffle(100) >>
build_batch >> network.train() >> plot >> Unzip())
print('train loss : {:.6f}'.format(t_loss >> Mean()))
print('train acc : {:.1f}'.format(100 * (t_acc >> Mean())))
e_acc = (test_samples >> transform >> build_batch >>
network.evaluate([categorical_accuracy]))
print('test acc : {:.1f}'.format(100 * e_acc))
network.save_best(e_acc, isloss=False)
"""
from __future__ import print_function
from glob import glob
from nutsflow import Collect, Consume, Get, Zip, Map, ArgMax, Print
from nutsml import (TransformImage, BuildBatch, ReadImage, ViewImageAnnotation,
ConvertLabel)
BATCH_SIZE = 128
if __name__ == "__main__":
from cnn_train import create_network, load_names
convert_label = ConvertLabel(None, load_names())
rerange = TransformImage(0).by('rerange', 0, 255, 0, 1, 'float32')
show_image = ViewImageAnnotation(0, 1, pause=1, figsize=(3, 3),
interpolation='spline36')
pred_batch = BuildBatch(BATCH_SIZE).input(0, 'image', 'float32')
print('loading network...')
network = create_network()
network.load_weights()
print('predicting...')
samples = glob('images/*.png') >> Print() >> ReadImage(None) >> Collect()
predictions = (samples >> rerange >> pred_batch >>
network.predict() >> convert_label)
samples >> Get(0) >> Zip(predictions) >> show_image >> Consume()
.. module:: mlp_predict
:synopsis: Example nuts-ml pipeline for classification
"""
from __future__ import print_function
from nutsflow import Collect, Consume, Get, Zip, Map, Format, ArgMax
from nutsml import (TransformImage, BuildBatch, ReadImage, ReadLabelDirs,
ViewImageAnnotation)
BATCH_SIZE = 128
if __name__ == "__main__":
from mlp_train import create_network
TransformImage.register('flatten', lambda img: img.flatten())
transform = (TransformImage(0).by('rerange', 0, 255, 0, 1,
'float32').by('flatten'))
show_image = ViewImageAnnotation(0, (1, 2), pause=1, figsize=(4, 4))
pred_batch = BuildBatch(BATCH_SIZE).by(0, 'vector', 'float32')
print('loading network...')
network = create_network()
network.load_weights()
print('predicting...')
samples = ReadLabelDirs('images', '*.png') >> ReadImage(0) >> Collect()
truelabels = samples >> Get(1) >> Format('true: {}')
predictions = (samples >> transform >> pred_batch >> network.predict() >>
Map(ArgMax()) >> Format('pred: {}'))
samples >> Get(0) >> Zip(predictions,
"""
.. module:: view_train_images
:synopsis: Example for showing images with transformation
"""
from nutsflow import Take, Consume, GetCols
from nutsml import ViewImage, TransformImage
if __name__ == "__main__":
from mlp_train import load_samples
samples, _ = load_samples()
transform = (TransformImage(0).by('elastic', 5, 100))
(samples >> GetCols(0, 0, 1) >> Take(1000) >> transform >> ViewImage(
(0, 1), pause=1) >> Consume())