opencv/samples/python/digits_adjust.py

140 lines
4.3 KiB
Python
Raw Normal View History

2013-03-06 14:41:02 +08:00
#!/usr/bin/env python
2012-10-17 07:18:30 +08:00
'''
Digit recognition adjustment.
Grid search is used to find the best parameters for SVM and KNearest classifiers.
SVM adjustment follows the guidelines given in
http://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf
Usage:
digits_adjust.py [--model {svm|knearest}]
2012-10-17 07:18:30 +08:00
--model {svm|knearest} - select the classifier (SVM is the default)
'''
# Python 2/3 compatibility
from __future__ import print_function
import sys
PY3 = sys.version_info[0] == 3
if PY3:
xrange = range
2012-10-17 07:18:30 +08:00
import numpy as np
import cv2 as cv
2012-10-17 07:18:30 +08:00
from multiprocessing.pool import ThreadPool
from digits import *
def cross_validate(model_class, params, samples, labels, kfold = 3, pool = None):
n = len(samples)
folds = np.array_split(np.arange(n), kfold)
def f(i):
model = model_class(**params)
test_idx = folds[i]
train_idx = list(folds)
train_idx.pop(i)
train_idx = np.hstack(train_idx)
train_samples, train_labels = samples[train_idx], labels[train_idx]
test_samples, test_labels = samples[test_idx], labels[test_idx]
model.train(train_samples, train_labels)
resp = model.predict(test_samples)
score = (resp != test_labels).mean()
print(".", end='')
2012-10-17 07:18:30 +08:00
return score
if pool is None:
scores = list(map(f, xrange(kfold)))
2012-10-17 07:18:30 +08:00
else:
scores = pool.map(f, xrange(kfold))
return np.mean(scores)
class App(object):
def __init__(self):
self._samples, self._labels = self.preprocess()
2012-10-17 07:18:30 +08:00
def preprocess(self):
digits, labels = load_digits(DIGITS_FN)
shuffle = np.random.permutation(len(digits))
digits, labels = digits[shuffle], labels[shuffle]
digits2 = list(map(deskew, digits))
2012-10-17 07:18:30 +08:00
samples = preprocess_hog(digits2)
return samples, labels
def get_dataset(self):
return self._samples, self._labels
2012-10-17 07:18:30 +08:00
def run_jobs(self, f, jobs):
pool = ThreadPool(processes=cv.getNumberOfCPUs())
ires = pool.imap_unordered(f, jobs)
2012-10-17 07:18:30 +08:00
return ires
def adjust_SVM(self):
Cs = np.logspace(0, 10, 15, base=2)
gammas = np.logspace(-7, 4, 15, base=2)
scores = np.zeros((len(Cs), len(gammas)))
scores[:] = np.nan
print('adjusting SVM (may take a long time) ...')
2012-10-17 07:18:30 +08:00
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)
2012-10-17 07:18:30 +08:00
print('writing score table to "svm_scores.npz"')
2012-10-17 07:18:30 +08:00
np.savez('svm_scores.npz', scores=scores, Cs=Cs, gammas=gammas)
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))
2012-10-17 07:18:30 +08:00
return best_params
def adjust_KNearest(self):
print('adjusting KNearest ...')
2012-10-17 07:18:30 +08:00
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))
2012-10-17 07:18:30 +08:00
best_params = dict(k=best_k)
print('best params:', best_params, 'err: %.2f' % (best_err*100))
2012-10-17 07:18:30 +08:00
return best_params
if __name__ == '__main__':
import getopt
import sys
print(__doc__)
2012-10-17 07:18:30 +08:00
args, _ = getopt.getopt(sys.argv[1:], '', ['model='])
2012-10-17 07:18:30 +08:00
args = dict(args)
args.setdefault('--model', 'svm')
args.setdefault('--env', '')
if args['--model'] not in ['svm', 'knearest']:
print('unknown model "%s"' % args['--model'])
2012-10-17 07:18:30 +08:00
sys.exit(1)
t = clock()
app = App()
2012-10-17 07:18:30 +08:00
if args['--model'] == 'knearest':
app.adjust_KNearest()
else:
app.adjust_SVM()
print('work time: %f s' % (clock() - t))