generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merged the trunk r8852:8880

Browse Source
This commit is contained in:
Andrey Kamaev
2012-07-02 11:04:43 +00:00
parent 304dac7f00
commit b368f99d03
60 changed files with 507 additions and 255 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
samples/python2/digits.py
View File

@@ -17,6 +17,7 @@ from common import clock, mosaic
SZ = 20 # size of each digit is SZ x SZ
CLASS_N = 10
DIGITS_FN = 'digits.png'
def load_digits(fn):
print 'loading "%s" ...' % fn
@@ -95,7 +96,7 @@ def evaluate_model(model, digits, samples, labels):
if __name__ == '__main__':
print __doc__
digits, labels = load_digits('digits.png')
digits, labels = load_digits(DIGITS_FN)
print 'preprocessing...'
# shuffle digits

156
samples/python2/digits_adjust.py
View File

@@ -11,11 +11,10 @@ Usage:
digits_adjust.py [--model {svm|knearest}] [--cloud] [--env <PiCloud environment>]
--model {svm|knearest} - select the classifier (SVM is the default)
--cloud - use PiCloud computing platform (for SVM only)
--cloud - use PiCloud computing platform
--env - cloud environment name
'''
# TODO dataset preprocessing in cloud
# TODO cloud env setup tutorial
import numpy as np
@@ -24,6 +23,14 @@ from multiprocessing.pool import ThreadPool
from digits import *
try:
import cloud
have_cloud = True
except ImportError:
have_cloud = False
def cross_validate(model_class, params, samples, labels, kfold = 3, pool = None):
n = len(samples)
folds = np.array_split(np.arange(n), kfold)
@@ -46,66 +53,88 @@ def cross_validate(model_class, params, samples, labels, kfold = 3, pool = None)
scores = pool.map(f, xrange(kfold))
return np.mean(scores)
def adjust_KNearest(samples, labels):
print 'adjusting KNearest ...'
best_err, best_k = np.inf, -1
for k in xrange(1, 9):
err = cross_validate(KNearest, dict(k=k), samples, labels)
if err < best_err:
best_err, best_k = err, k
print 'k = %d, error: %.2f %%' % (k, err*100)
best_params = dict(k=best_k)
print 'best params:', best_params
return best_params
def adjust_SVM(samples, labels, usecloud=False, cloud_env=''):
Cs = np.logspace(0, 5, 10, base=2)
gammas = np.logspace(-7, -2, 10, base=2)
scores = np.zeros((len(Cs), len(gammas)))
scores[:] = np.nan
if usecloud:
try:
import cloud
except ImportError:
print 'cloud module is not installed'
class App(object):
def __init__(self, usecloud=False, cloud_env=''):
if usecloud and not have_cloud:
print 'warning: cloud module is not installed, running locally'
usecloud = False
if usecloud:
print 'uploading dataset to cloud...'
np.savez('train.npz', samples=samples, labels=labels)
cloud.files.put('train.npz')
self.usecloud = usecloud
self.cloud_env = cloud_env
print 'adjusting SVM (may take a long time) ...'
def f(job):
i, j = job
params = dict(C = Cs[i], gamma=gammas[j])
score = cross_validate(SVM, params, samples, labels)
return i, j, score
def fcloud(job):
i, j = job
cloud.files.get('train.npz')
npz = np.load('train.npz')
params = dict(C = Cs[i], gamma=gammas[j])
score = cross_validate(SVM, params, npz['samples'], npz['labels'])
return i, j, score
if usecloud:
jids = cloud.map(fcloud, np.ndindex(*scores.shape), _env=cloud_env, _profile=True)
ires = cloud.iresult(jids)
else:
pool = ThreadPool(processes=cv2.getNumberOfCPUs())
ires = pool.imap_unordered(f, np.ndindex(*scores.shape))
if self.usecloud:
print 'uploading dataset to cloud...'
cloud.files.put(DIGITS_FN)
self.preprocess_job = cloud.call(self.preprocess, _env=self.cloud_env)
else:
self._samples, self._labels = self.preprocess()
for count, (i, j, score) in enumerate(ires):
scores[i, j] = score
print '%d / %d (best error: %.2f %%, last: %.2f %%)' % (count+1, scores.size, np.nanmin(scores)*100, score*100)
print scores
def preprocess(self):
if self.usecloud:
cloud.files.get(DIGITS_FN)
digits, labels = load_digits(DIGITS_FN)
shuffle = np.random.permutation(len(digits))
digits, labels = digits[shuffle], labels[shuffle]
digits2 = map(deskew, digits)
samples = np.float32(digits2).reshape(-1, SZ*SZ) / 255.0
return samples, labels
def get_dataset(self):
if self.usecloud:
return cloud.result(self.preprocess_job)
else:
return self._samples, self._labels
def run_jobs(self, f, jobs):
if self.usecloud:
jids = cloud.map(f, jobs, _env=self.cloud_env, _profile=True, _depends_on=self.preprocess_job)
ires = cloud.iresult(jids)
else:
pool = ThreadPool(processes=cv2.getNumberOfCPUs())
ires = pool.imap_unordered(f, jobs)
return ires
def adjust_SVM(self):
Cs = np.logspace(0, 5, 10, base=2)
gammas = np.logspace(-7, -2, 10, base=2)
scores = np.zeros((len(Cs), len(gammas)))
scores[:] = np.nan
print 'adjusting SVM (may take a long time) ...'
def f(job):
i, j = job
samples, labels = self.get_dataset()
params = dict(C = Cs[i], gamma=gammas[j])
score = cross_validate(SVM, params, samples, labels)
return i, j, score
ires = self.run_jobs(f, np.ndindex(*scores.shape))
for count, (i, j, score) in enumerate(ires):
scores[i, j] = score
print '%d / %d (best error: %.2f %%, last: %.2f %%)' % (count+1, scores.size, np.nanmin(scores)*100, score*100)
print scores
i, j = np.unravel_index(scores.argmin(), scores.shape)
best_params = dict(C = Cs[i], gamma=gammas[j])
print 'best params:', best_params
print 'best error: %.2f %%' % (scores.min()*100)
return best_params
def adjust_KNearest(self):
print 'adjusting KNearest ...'
def f(k):
samples, labels = self.get_dataset()
err = cross_validate(KNearest, dict(k=k), samples, labels)
return k, err
best_err, best_k = np.inf, -1
for k, err in self.run_jobs(f, xrange(1, 9)):
if err < best_err:
best_err, best_k = err, k
print 'k = %d, error: %.2f %%' % (k, err*100)
best_params = dict(k=best_k)
print 'best params:', best_params, 'err: %.2f' % (best_err*100)
return best_params
i, j = np.unravel_index(scores.argmin(), scores.shape)
best_params = dict(C = Cs[i], gamma=gammas[j])
print 'best params:', best_params
print 'best error: %.2f %%' % (scores.min()*100)
return best_params
if __name__ == '__main__':
import getopt
@@ -113,6 +142,7 @@ if __name__ == '__main__':
print __doc__
args, _ = getopt.getopt(sys.argv[1:], '', ['model=', 'cloud', 'env='])
args = dict(args)
args.setdefault('--model', 'svm')
@@ -121,16 +151,10 @@ if __name__ == '__main__':
print 'unknown model "%s"' % args['--model']
sys.exit(1)
digits, labels = load_digits('digits.png')
shuffle = np.random.permutation(len(digits))
digits, labels = digits[shuffle], labels[shuffle]
digits2 = map(deskew, digits)
samples = np.float32(digits2).reshape(-1, SZ*SZ) / 255.0
t = clock()
app = App(usecloud='--cloud' in args, cloud_env = args['--env'])
if args['--model'] == 'knearest':
adjust_KNearest(samples, labels)
app.adjust_KNearest()
else:
adjust_SVM(samples, labels, usecloud='--cloud' in args, cloud_env = args['--env'])
app.adjust_SVM()
print 'work time: %f s' % (clock() - t)

78
samples/python2/fitline.py Normal file
View File

@@ -0,0 +1,78 @@
'''
Robust line fitting.
==================
Example of using cv2.fitLine function for fitting line to points in presence of outliers.
Usage
-----
fitline.py
Switch through different M-estimator functions and see, how well the robust functions
fit the line even in case of ~50% of outliers.
Keys
----
SPACE - generaty random points
f - change distance function
ESC - exit
'''
import numpy as np
import cv2
import itertools as it
from common import draw_str
w, h = 512, 256
def toint(p):
return tuple(map(int, p))
def sample_line(p1, p2, n, noise=0.0):
p1 = np.float32(p1)
t = np.random.rand(n,1)
return p1 + (p2-p1)*t + np.random.normal(size=(n, 2))*noise
dist_func_names = it.cycle('CV_DIST_L2 CV_DIST_L1 CV_DIST_L12 CV_DIST_FAIR CV_DIST_WELSCH CV_DIST_HUBER'.split())
cur_func_name = dist_func_names.next()
def update(_=None):
noise = cv2.getTrackbarPos('noise', 'fit line')
n = cv2.getTrackbarPos('point n', 'fit line')
r = cv2.getTrackbarPos('outlier %', 'fit line') / 100.0
outn = int(n*r)
p0, p1 = (90, 80), (w-90, h-80)
img = np.zeros((h, w, 3), np.uint8)
cv2.line(img, toint(p0), toint(p1), (0, 255, 0))
if n > 0:
line_points = sample_line(p0, p1, n-outn, noise)
outliers = np.random.rand(outn, 2) * (w, h)
points = np.vstack([line_points, outliers])
for p in line_points:
cv2.circle(img, toint(p), 2, (255, 255, 255), -1)
for p in outliers:
cv2.circle(img, toint(p), 2, (64, 64, 255), -1)
func = getattr(cv2.cv, cur_func_name)
vx, vy, cx, cy = cv2.fitLine(np.float32(points), func, 0, 0.01, 0.01)
cv2.line(img, (int(cx-vx*w), int(cy-vy*w)), (int(cx+vx*w), int(cy+vy*w)), (0, 0, 255))
draw_str(img, (20, 20), cur_func_name)
cv2.imshow('fit line', img)
if __name__ == '__main__':
print __doc__
cv2.namedWindow('fit line')
cv2.createTrackbar('noise', 'fit line', 3, 50, update)
cv2.createTrackbar('point n', 'fit line', 100, 500, update)
cv2.createTrackbar('outlier %', 'fit line', 30, 100, update)
while True:
update()
ch = cv2.waitKey(0)
if ch == ord('f'):
cur_func_name = dist_func_names.next()
if ch == 27:
break

38
samples/python2/video.py
View File

@@ -1,3 +1,32 @@
'''
Video capture sample.
Sample shows how VideoCapture class can be used to acquire video
frames from a camera of a movie file. Also the sample provides
an example of procedural video generation by an object, mimicking
the VideoCapture interface (see Chess class).
'create_capture' is a convinience function for capture creation,
falling back to procedural video in case of error.
Usage:
video.py [--shotdir <shot path>] [source0] [source1] ...'
sourceN is an
- integer number for camera capture
- name of video file
- synth:<params> for procedural video
Synth examples:
synth:bg=../cpp/lena.jpg:noise=0.1
synth:class=chess:bg=../cpp/lena.jpg:noise=0.1:size=640x480
Keys:
ESC - exit
SPACE - save current frame to <shot path> directory
'''
import numpy as np
import cv2
from time import clock
@@ -100,8 +129,7 @@ presets = dict(
def create_capture(source = 0, fallback = presets['chess']):
'''
source: <int> or '<int>|<filename>|synth [:<param_name>=<value> [:...]]'
'''source: <int> or '<int>|<filename>|synth [:<param_name>=<value> [:...]]'
'''
source = str(source).strip()
chunks = source.split(':')
@@ -136,9 +164,7 @@ if __name__ == '__main__':
import sys
import getopt
print 'USAGE: video.py [--shotdir <dir>] [source0] [source1] ...'
print "source: '<int>' or '<filename>' or 'synth:<params>'"
print
print __doc__
args, sources = getopt.getopt(sys.argv[1:], '', 'shotdir=')
args = dict(args)
@@ -146,8 +172,6 @@ if __name__ == '__main__':
if len(sources) == 0:
sources = [ 0 ]
print 'Press SPACE to save current frame'
caps = map(create_capture, sources)
shot_idx = 0
while True:

29
samples/python2/watershed.py
View File

@@ -1,18 +1,31 @@
import numpy as np
import cv2
from common import Sketcher
'''
Watershed segmentation
=========
help_message = '''
USAGE: watershed.py [<image>]
This program demonstrates the watershed segmentation algorithm
in OpenCV: watershed().
Use keys 1 - 7 to switch marker color
Usage
-----
watershed.py [image filename]
Keys
----
1-7 - switch marker color
SPACE - update segmentation
r - reset
a - switch autoupdate
a - toggle autoupdate
ESC - exit
'''
import numpy as np
import cv2
from common import Sketcher
class App:
def __init__(self, fn):
self.img = cv2.imread(fn)
@@ -60,5 +73,5 @@ if __name__ == '__main__':
import sys
try: fn = sys.argv[1]
except: fn = '../cpp/fruits.jpg'
print help_message
print __doc__
App(fn).run()