numParts = args.numParts
numExtensionParts = args.numExtensionParts
extensionPatchSize = args.extensionPatchSize
if args.log:
from pnet.vzlog import default as vz
ag.set_verbose(True)
sup_ims = []
sup_labels = []
net = pnet.PartsNet.load(args.lowerPartsModel)
layers = [
#pnet.IntensityThresholdLayer(),
pnet.ExtensionPartsLayer(num_parts = numParts, num_components = numExtensionParts, part_shape = (extensionPatchSize, extensionPatchSize), lowerLayerShape = (6,6))
]
clnet = pnet.PartsNet([net] + layers)
ims, label = ag.io.load_cifar10('training')
clnet.train(ims)
clnet.save(args.model)
'''
classes = range(10)
classificationTraining = 5000
rs = np.random.RandomState(training_seed)
for d in classes:
ims0,tmpLabel = ag.io.load_cifar10('training', classes = [d])
ag.set_verbose(True)
sup_ims = []
sup_labels = []
net = None
layers = [
#pnet.IntensityThresholdLayer(),
pnet.EdgeLayer(k=5, radius=1, spread='orthogonal', minimum_contrast=0.05),
pnet.PartsLayer(100, (6, 6), settings=dict(outer_frame=0,
threshold=40,
samples_per_image=40,
max_samples=1000000,
min_prob=0.005,
)),
pnet.ExtensionPartsLayer(num_parts = 100, num_components = 10, part_shape = (12,12), lowerLayerShape = (6,6)),
pnet.PoolingLayer(shape=(4,4), strides=(4, 4)),
pnet.MixtureClassificationLayer(n_components=1, min_prob=0.0001,block_size=200),
#pnet.SVMClassificationLayer(C=None)
]
net = pnet.PartsNet(layers)
digits = range(10)
ims = ag.io.load_mnist('training', selection=slice(10000), return_labels=False)
net.train(ims)
#sup_ims = []
#sup_labels = []
radius=1,
spread='orthogonal',
minimum_contrast=0.05),
pnet.PartsLayer(100, (3, 3),
settings=dict(
outer_frame=0,
threshold=10,
samples_per_image=20,
max_samples=1000000,
min_prob=0.005,
)),
#pnet.PoolingLayer(shape=(4,4), strides=(4, 4)),
#pnet.MixtureClassificationLayer(n_components=1, min_prob=0.0001,block_size=200),
pnet.ExtensionPartsLayer(num_parts=100,
num_components=10,
part_shape=(9, 9),
lowerLayerShape=(3, 3)),
#pnet.ExtensionPoolingLayer(n_parts = 1000, grouping_type = 'rbm', pooling_type = 'distance', pooling_distance = 5, weights_file = None, save_weights_file = None, settings = {})
pnet.PoolingLayer(shape=(4, 4), strides=(4, 4)),
pnet.SVMClassificationLayer(C=None)
]
net = pnet.PartsNet(layers)
ims, label = ag.io.load_cifar10('training', selection=slice(0, 1000))
net.train(ims)
classes = range(10)
classificationTraining = 10
rs = np.random.RandomState(training_seed)