def main(folder):
""" main routine of the program """
# parse the command line arguments
description = 'Average ground profiles from several analyzed videos'
parser = argparse.ArgumentParser(description=description)
parser.add_argument('-o', '--output', dest='output', type=str,
help='file to which the output statistics are written')
parser.add_argument('-p', '--plot', dest='plot',
action='store_true', help='plots the result')
parser.add_argument('files', metavar='file', type=str, nargs='+',
help='files to analyze')
args = parser.parse_args()
# get files to analyze
filenames = args.files
# analyze all the files
parameters = {'logging/enabled': False}
ground, traces = average_ground_shape(filenames, ret_traces=True,
parameters=parameters)
if args.plot:
# show the result if requested
plt.plot(ground[:, 0], ground[:, 1], 'b-', lw=3)
# determine traces that are far from the average
dist = []
for k, trace in enumerate(traces):
ys = np.interp(ground[:, 0], trace[:, 0], trace[:, 1])
dist.append(np.linalg.norm(ground[:, 1] - ys))
if dist[-1] < 0.3:
plt.plot(trace[:, 0], trace[:, 1], 'k-', alpha=0.2)
if dist[-1] > 0.2:
print 'Possibly bad ground profile', filenames[k]
else:
plt.plot(trace[:, 0], trace[:, 1], 'r-', alpha=0.2)
print 'Bad ground profile', filenames[k]
traces = [t for k, t in enumerate(traces)
if dist[k] < 0.3]
ground = curves.average_normalized_functions(traces)
plt.plot(ground[:, 0], ground[:, 1], 'g-', lw=3)
plt.gca().invert_yaxis()
plt.show()
if args.output:
# write to file
ext = os.path.splitext(args.output)[1]
if ext == '.pkl':
pickle.dump(ground, open(args.output, 'w'))
elif ext == '.yaml':
yaml.dump(ground.tolist(), open(args.output, 'w'))
else:
raise ValueError('Unknown output format `%s`' % ext[1:])
else:
# write to stdout
print ground.tolist()
def average_ground_shape(result_files, time_point=60*60, parameters=None,
ret_traces=False):
""" determines an average ground shape from a list of results """
# get all the profiles
profiles = []
for filename in result_files:
# load the results
try:
result = load_result_file(filename, parameters)
ground_profile = result.data['pass2/ground_profile']
except KeyError:
logging.warn('Data of `%s` could not be read', filename)
# retrieve profile at the right time point
if result.use_units:
frame_id = time_point/(result.time_scale/result.units.second)
else:
frame_id = time_point/result.time_scale
profile = ground_profile.get_ground_profile(frame_id)
# scale profile such that width=1
points = profile.line
scale = points[-1][0] - points[0][0]
points[:, 0] = (points[:, 0] - points[0, 0])/scale
points[:, 1] = (points[:, 1] - points[:, 1].mean())/scale
profiles.append(points)
# average the profiles
profile_avg = curves.average_normalized_functions(profiles)
if ret_traces:
return profile_avg, profiles
else:
return profile_avg
def average_ground_shape(result_files,
time_point=60 * 60,
parameters=None,
ret_traces=False):
""" determines an average ground shape from a list of results """
# get all the profiles
profiles = []
for filename in result_files:
# load the results
try:
result = load_result_file(filename, parameters)
ground_profile = result.data['pass2/ground_profile']
except KeyError:
logging.warn('Data of `%s` could not be read', filename)
# retrieve profile at the right time point
if result.use_units:
frame_id = time_point / (result.time_scale / result.units.second)
else:
frame_id = time_point / result.time_scale
profile = ground_profile.get_ground_profile(frame_id)
# scale profile such that width=1
points = profile.line
scale = points[-1][0] - points[0][0]
points[:, 0] = (points[:, 0] - points[0, 0]) / scale
points[:, 1] = (points[:, 1] - points[:, 1].mean()) / scale
profiles.append(points)
# average the profiles
profile_avg = curves.average_normalized_functions(profiles)
if ret_traces:
return profile_avg, profiles
else:
return profile_avg
def main(folder):
""" main routine of the program """
# parse the command line arguments
description = 'Average ground profiles from several analyzed videos'
parser = argparse.ArgumentParser(description=description)
parser.add_argument('-o',
'--output',
dest='output',
type=str,
help='file to which the output statistics are written')
parser.add_argument('-p',
'--plot',
dest='plot',
action='store_true',
help='plots the result')
parser.add_argument('files',
metavar='file',
type=str,
nargs='+',
help='files to analyze')
args = parser.parse_args()
# get files to analyze
filenames = args.files
# analyze all the files
parameters = {'logging/enabled': False}
ground, traces = average_ground_shape(filenames,
ret_traces=True,
parameters=parameters)
if args.plot:
# show the result if requested
plt.plot(ground[:, 0], ground[:, 1], 'b-', lw=3)
# determine traces that are far from the average
dist = []
for k, trace in enumerate(traces):
ys = np.interp(ground[:, 0], trace[:, 0], trace[:, 1])
dist.append(np.linalg.norm(ground[:, 1] - ys))
if dist[-1] < 0.3:
plt.plot(trace[:, 0], trace[:, 1], 'k-', alpha=0.2)
if dist[-1] > 0.2:
print 'Possibly bad ground profile', filenames[k]
else:
plt.plot(trace[:, 0], trace[:, 1], 'r-', alpha=0.2)
print 'Bad ground profile', filenames[k]
traces = [t for k, t in enumerate(traces) if dist[k] < 0.3]
ground = curves.average_normalized_functions(traces)
plt.plot(ground[:, 0], ground[:, 1], 'g-', lw=3)
plt.gca().invert_yaxis()
plt.show()
if args.output:
# write to file
ext = os.path.splitext(args.output)[1]
if ext == '.pkl':
pickle.dump(ground, open(args.output, 'w'))
elif ext == '.yaml':
yaml.dump(ground.tolist(), open(args.output, 'w'))
else:
raise ValueError('Unknown output format `%s`' % ext[1:])
else:
# write to stdout
print ground.tolist()
