def get_snre( window_types, snre_record, reg, filename_creator ):
rec = pylab.csv2rec( filename_creator('blackman'))
strains = list( set( rec['strain_percent'] ) )
strains.sort()
index_is_reg = rec['regularization'] == reg
mean_snre = np.ones( (len( window_types ), len( strains ))) * 10**-5
snre_error = np.zeros( (len( window_types ), len( strains )))
for ri, rad in enumerate( window_types ):
rec = pylab.csv2rec( filename_creator(rad))
strains = list( set( rec['strain_percent'] ) )
strains.sort()
index_is_reg = rec['regularization'] == reg
for si, strain in enumerate( strains ):
index_is_strain = rec['strain_percent'] == strain
strain_indices = np.arange( len( rec ) )[ np.nonzero(
np.logical_and( index_is_strain, index_is_reg ))]
if snre_record.find( 'lateral' ) == -1:
strains[si] = strain
else:
strains[si] = strain / 2.0
trials = rec[snre_record][strain_indices]
if len( trials ) > 0:
mean_snre[ri, si] = np.mean( trials )
snre_error[ri,si] = 2*np.std( trials )/np.sqrt( len( trials ))
return strains, mean_snre, snre_error
def read_listfile(filename):
"""
Reads in data sample list stored as a CSV file with optional meta-data.
Parameters:
-----------
filename: string
Input filename.
Returns:
--------
meta: dict
Meta-data from the header.
data: recarray
Data from the CSV table.
"""
if filename.endswith('.gz'):
open_fn = gzip.GzipFile
else:
open_fn = open
with open_fn(filename, 'r') as f:
meta = [ line[1:] for line in f.readlines() if line.startswith('#') ]
meta = yaml.load(''.join(meta))
data = csv2rec(filename, delimiter='\t', comments='#')
assert 'id' in data.dtype.names
assert len(data) == len(np.unique(data['id']))
return meta, data
def draw3d(filename):
imSimCRec = pylab.csv2rec(filename)
imSimCArray = np.array(imSimCRec.tolist())
fig1 = plt.figure()
ax1 = fig1.gca(projection='3d')
X = np.arange(20, 170, 1)
Y = np.arange(20, 170, 1)
X, Y = np.meshgrid(X, Y)
ax1.plot_surface(X, Y, imSimCArray, cmap=cm.coolwarm)# rstride=8, cstride=8, alpha=0.3)
ax1.set_xlabel('X')
ax1.set_xlim(20, 170)
ax1.set_ylabel('Y')
ax1.set_ylim(20, 170)
markeredgewidth=1.0,
markevery=1,
ms=9.0,
alpha=0.5,
)
plt.xlabel("Strain Percent Magnitude")
plt.xlim(0.0, 8.0)
plt.ylabel(ylabel)
# plt.title( str( variable_name ))
plt.legend(loc="best")
if logplot:
plt.gca().set_yscale("log")
plt.ylim((10 ** -2, 10 ** 2))
r = pylab.csv2rec(sys.argv[1])
variable_name = r.dtype.names[0]
# I don't think this is actually necessary the way I ended up doing it.
sorted_indices = np.lexsort((r["strain_percent"], r[variable_name], r["regularization"]))
regs = set(r["regularization"])
for reg in regs:
print(reg)
index_is_reg = r["regularization"][sorted_indices] == reg
plot_results = PlotResults(r, sorted_indices)
print("axial strain SNRe")
plot_results.plot_snre("axial_strain_snre", "Axial $SNR_e$ [dB]", index_is_reg)
print("lateral strain SNRe")
plot_results.plot_snre("lateral_strain_snre", "lateral $SNR_e$ " + str(reg), index_is_reg)
# coding: utf-8
# In[19]:
import numpy as np
import matplotlib.pylab as plt
import matplotlib.dates as mdates
# import the csv to a numpy array
datas = plt.csv2rec("programmingLanguageComparaison.csv")
# Set the figure size big to see it clearly
plt.figure(figsize=(20, 10))
dates = datas["week"]
# Init the trendsNames for the legend
trendsNames = ["python", "javascript", "C++", "SQL"]
# Init the trends datas to plot them
trendsDatas = [datas["python"], datas["javascript"], datas["c"], datas["sql"]]
# Init the colors
trendsColors = ["#71D4FF", "white", "#FCCF4D", "#FC5757"]
# print(datas["sql"])
# For each trends : [trendName, maxValue, maxValueDate, minValue, minValueDate]
peaks = []
for i in range(len(trendsDatas)):
# create a temporary list
def is_touching(coarse_segmentation, dirname, ids):
d = os.path.join(coarse_segmentation, dirname)
objects = csv2rec(os.path.join(d, 'objects.csv'), delimiter=';')
components = csv2rec(os.path.join(d, 'components.csv'), delimiter=';')
return [ components[components['bbox'] == objects[objects['obj'] == int(metasys_id)][0]['bbox']]['cut'].any()
for metasys_id in ids ]
for metasys_id in ids ]
def is_touching(coarse_segmentation, dirname, ids):
d = os.path.join(coarse_segmentation, dirname)
objects = csv2rec(os.path.join(d, 'objects.csv'), delimiter=';')
components = csv2rec(os.path.join(d, 'components.csv'), delimiter=';')
return [ components[components['bbox'] == objects[objects['obj'] == int(metasys_id)][0]['bbox']]['cut'].any()
for metasys_id in ids ]
fine_segmentation_dir = '/groups/stark/projects/vt/Segmentation/Fine/MetaSys/'
coarse_segmentation_dir = '/groups/stark/projects/vt/Segmentation/Coarse/MetaSys/'
if __name__ == '__main__':
import sys
filename = sys.argv[1]
data = csv2rec(filename, delimiter=',')
dirname = data['dirname'][0]
ids = data['metasys_id']
blurred = is_blurred(fine_segmentation_dir, dirname, ids)
touching = is_touching(coarse_segmentation_dir, dirname, ids)
import pandas
df = pandas.DataFrame(data)
df['blurred'] = blurred
df['touching'] = touching
df['bad_outline'] = df['blurred'] + df['touching']
outname = os.path.splitext(filename)
outname = outname[0] + '-wb.csv'
rec2csv(df.to_records(), outname, delimiter=',')
#plt.title( str( variable_name ))
plt.legend( loc='best' )
if self.logplot:
plt.gca().set_yscale( 'log' )
plt.ylim( (10**-1.2, 10**1.5 ) )
plt.xlim( (0.0, s[-1]+0.5) )
#plt.ylim( (0.0, 6.0 ) )
plot_results = PlotResults()
curves = [ ('noInterp_sim', 'No Interpolation'),
('cosine_sim', 'Cosine'),
('parabolic_sim', 'Parabolic'),
('sinc_sim', 'Sinc - Amoeba'),
('gradientDescent_sim', 'Sinc - Gradient Descent') ]
for curve, curve_name in curves:
r = pylab.csv2rec( 'SNRe_' + curve + '.csv' )
plot_results.plot_snre_curves( curve, curve_name, r )
plt.figure( 1 )
plt.savefig( '../../../../doc/images/interp_method_simulation_regularization_axial.png' )
plt.savefig( '../../../../doc/images/interp_method_simulation_regularization_axial.eps' )
plt.figure( 2 )
plt.savefig( '../../../../doc/images/interp_method_simulation_regularization_lateral.png' )
plt.savefig( '../../../../doc/images/interp_method_simulation_regularization_lateral.eps' )
plt.figure( 3 )
plt.savefig( '../../../../doc/images/interp_method_simulation_no_regularization_axial.png' )
plt.savefig( '../../../../doc/images/interp_method_simulation_no_regularization_axial.eps' )
plt.figure( 4 )
plt.savefig( '../../../../doc/images/interp_method_simulation_no_regularization_lateral.png' )
plt.savefig( '../../../../doc/images/interp_method_simulation_no_regularization_lateral.eps' )
#! /usr/bin/env python
import sys, argparse, datetime
import storage, cvutils, utils
import matplotlib.pylab as pylab
import matplotlib.pyplot as plt
import numpy as np
annotations = pylab.csv2rec(sys.argv[1])
frameRate = 30
dirname = "/home/nicolas/Research/Data/montreal/infractions-pietons/"
videoDirnames = {'amherst': '2011-06-22-sherbrooke-amherst/',
'iberville': '2011-06-28-sherbrooke-iberville/'}
# for amherst, subtract 40 seconds: add a delta
for annotation in annotations:
video = annotation['video_name'].lower()
print('{} {}'.format(annotation['conflict_start_time'], annotation['conflict_end_time']))
print(annotation['road_user_1']+' '+annotation['road_user_2']+' '+annotation['conflict_quality'])
print(annotation['comments'])
cvutils.playVideo(dirname+videoDirnames[video]+video+'-{}.avi'.format(annotation['video_start_time']), utils.timeToFrames(annotation['conflict_start_time']+datetime.timedelta(seconds=-40), frameRate), frameRate, True, False, annotation['road_user_1']+' '+annotation['road_user_2']+' '+annotation['conflict_quality'])
labels = truth_labels
del truth_labels[-1]
dtype += [(opts.truth, int)]
rescan_map = {'rescan': 1, 'rescan2': 2, 'rescan3': 3}
dst = np.array([], dtype=dtype)
t = -1
dirnames = []
csvnames = sorted(glob.glob(csv_pattern))
if len(csvnames) == 0:
raise RuntimeError('No input files matching the pattern.')
for csvname in csvnames:
label = os.path.splitext(os.path.basename(csvname))[0]
if label.find('#') != -1:
continue
t = max(t, os.path.getmtime(csvname))
src = csv2rec(csvname, delimiter=',', converterd=converterd)
for sample in src:
sample_dirname = sample['dirname'].split('/')
sample_date = sample_dirname[-2].split('-')
if len(sample_date) > 1:
sample_date = int(
sample_date[0]) * 10 + rescan_map[sample_date[1]]
else:
sample_date = int(sample_date[0]) * 10
sample_vt = sample_dirname[-1].split('~')
if len(sample_vt) > 1:
sample_suffix = ord(sample_vt[1]) - ord('A') + 1
sample_vt = int(sample_vt[0])
else:
sample_suffix = 0
sample_vt = int(sample_vt[0])
#! /usr/bin/env python
import sys, argparse, datetime
import storage, cvutils, utils
import matplotlib.pylab as pylab
import matplotlib.pyplot as plt
import numpy as np
annotations = pylab.csv2rec(sys.argv[1])
frameRate = 30
dirname = "/home/nicolas/Research/Data/montreal/infractions-pietons/"
videoDirnames = {
'amherst': '2011-06-22-sherbrooke-amherst/',
'iberville': '2011-06-28-sherbrooke-iberville/'
}
# for amherst, subtract 40 seconds: add a delta
for annotation in annotations:
video = annotation['video_name'].lower()
print('{} {}'.format(annotation['conflict_start_time'],
annotation['conflict_end_time']))
print(annotation['road_user_1'] + ' ' + annotation['road_user_2'] + ' ' +
annotation['conflict_quality'])
print(annotation['comments'])
cvutils.playVideo(
dirname + videoDirnames[video] + video +
'-{}.avi'.format(annotation['video_start_time']),
labels = truth_labels
del truth_labels[-1]
dtype += [(opts.truth, int)]
rescan_map = { 'rescan': 1, 'rescan2': 2, 'rescan3': 3 }
dst = np.array([], dtype=dtype)
t = -1
dirnames = []
csvnames = sorted(glob.glob(csv_pattern))
if len(csvnames) == 0:
raise RuntimeError('No input files matching the pattern.')
for csvname in csvnames:
label = os.path.splitext(os.path.basename(csvname))[0]
if label.find('#') != -1:
continue
t = max(t, os.path.getmtime(csvname))
src = csv2rec(csvname, delimiter=',', converterd=converterd)
for sample in src:
sample_dirname = sample['dirname'].split('/')
sample_date = sample_dirname[-2].split('-')
if len(sample_date) > 1:
sample_date = int(sample_date[0]) * 10 + rescan_map[sample_date[1]]
else:
sample_date = int(sample_date[0]) * 10
sample_vt = sample_dirname[-1].split('~')
if len(sample_vt) > 1:
sample_suffix = ord(sample_vt[1]) - ord('A') + 1
sample_vt = int(sample_vt[0])
else:
sample_suffix = 0
sample_vt = int(sample_vt[0])
sample_id = int(sample['metasys_id']) + 10000 * (
def draw2d(filename):
imSimHRec = pylab.csv2rec(filename)
imSimHArray = np.array(imSimHRec.tolist())
fig2 = plt.figure()
im2 = plt.imshow(imSimHArray, interpolation="none", )