def test_aegan(aegan, prefix):
import ipdb
with ipdb.launch_ipdb_on_exception():
aegan.load(prefix=prefix)
from GAN.utils.vis import vis_grid
vis_grid(inverse_transform(aegan.generator.random_generate(128)), (2, 20), 'random_generate.png')
paths = map(lambda x: x.strip(), open('protocol/cuhk01-all.txt').readlines())
from load import load_image
sample = transform( np.array([load_image(path, (64, 128)) for path in paths[:128]]) )
vis_grid(inverse_transform(sample), (2, 20), 'sample.png')
vis_grid(inverse_transform(aegan.autoencoder.autoencoder.predict(sample)), (2, 20), 'reconstruct.png')
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
codes = aegan.autoencoder.encoder.predict(sample)
# codes = aegan.generator.sample(128)
# codes = aegan.autoencoder.encoder.predict(aegan.generator.random_generate(128))
for ind, code in enumerate(codes):
n, bins, patches = plt.hist(code, 50, normed=1, facecolor='green', alpha=0.75)
plt.savefig('test/{}.pdf'.format(ind))
plt.clf()
ipdb.set_trace()
def feature_aegan(aegan, modelname, protoname):
with ipdb.launch_ipdb_on_exception():
aegan.load(prefix=modelname)
x = transform(load_all(protoname, (npxw, npxh)))
code = aegan.autoencoder.encoder.predict(x)
ipdb.set_trace()
def feature(aegan, filename):
import ipdb
with ipdb.launch_ipdb_on_exception():
aegan.load(prefix='./samples/reid_aegan/aegan/50')
paths = map(lambda x: x.strip(),
open('protocol/cuhk01-all.txt').readlines())
x = transform(np.array([load_image(path, (64, 128))
for path in paths]))
code = aegan.autoencoder.encoder.predict(x)
ipdb.set_trace()
plt.savefig('test/{}.pdf'.format(ind))
plt.clf()
ipdb.set_trace()
if __name__ == '__main__':
nbatch = 128
nmax = nbatch * 100
npxw, npxh = 64, 128
from load import people, load_all
va_data, tr_stream, _ = people(pathfile='protocol/cuhk01-train.txt',
size=(npxw, npxh),
batch_size=nbatch)
allx = transform(load_all('protocol/cuhk01-train.txt', (npxw, npxh)))
g = Generator(g_size=(3, npxh, npxw),
g_nb_filters=128,
g_nb_coding=5000,
g_scales=4,
g_init=InitNormal(scale=0.002)) #, g_FC=[5000])
d = Discriminator(d_size=g.g_size,
d_nb_filters=128,
d_scales=4,
d_init=InitNormal(scale=0.002)) #, d_FC=[5000])
# init with autoencoder
ae = Autoencoder(g, d)
# ae.fit(tr_stream,
# save_dir='./samples/reid_aegan_5000/ae/',