import vidfeat
import imfeat
import sklearn.svm
from vidfeat.models.synthetic_bovw_clusters import clusters
#import kernels
HOG = imfeat.HOGLatent(8, 2)
class FenceFrameFeature(vidfeat.ClassifierFrameFeature):
#feature = imfeat.MetaFeature(imfeat.GradientHistogram(), imfeat.Histogram('lab'))
feature = imfeat.MetaFeature(imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3))
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(FenceFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('fence', vidfeat.FenceFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
from vidfeat.models.screenshot_bovw_clusters import clusters
#import kernels
HOG = imfeat.HOGLatent(8, 2)
class ScreenshotFrameFeature(vidfeat.ClassifierFrameFeature):
#feature = imfeat.MetaFeature(imfeat.GradientHistogram(), imfeat.Histogram('lab'))
feature = imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(ScreenshotFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('screenshot', vidfeat.ScreenshotFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
class BlankFrameFeature(vidfeat.ClassifierFrameFeature):
feature = imfeat.BlackBars()
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(BlankFrameFeature, self).__init__(classifier=classifier, *args, **kw)
def _feature(self, image):
return self.feature(image)
def predict(self, frame):
if not frame.size:
return 0 if self._invert else 1
else:
return super(BlankFrameFeature, self).predict(frame)
if __name__ == '__main__':
vidfeat._frame_feature_main('blank', vidfeat.BlankFrameFeature, remove_bars=False)
import vidfeat
import imfeat
import sklearn.svm
from vidfeat.models.snow_bovw_clusters import clusters
HOG = imfeat.HOGLatent(8, 2)
class RoadFrameFeature(vidfeat.ClassifierFrameFeature):
#feature = imfeat.PyramidHistogram('lab')
feature = imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(RoadFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('road', vidfeat.RoadFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
from vidfeat.models.person_bovw_clusters import clusters
HOG = imfeat.HOGLatent(8, 2)
class PersonFrameFeature(vidfeat.ClassifierFrameFeature):
#feature = imfeat.PyramidHistogram('lab')
feature = imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(PersonFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('person', vidfeat.PersonFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
class UnderwaterFrameFeature(vidfeat.ClassifierFrameFeature):
feature = imfeat.MetaFeature(imfeat.Histogram('lab'), imfeat.GIST(), max_side=128)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(UnderwaterFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
return self.feature(image)
if __name__ == '__main__':
vidfeat._frame_feature_main('underwater', vidfeat.UnderwaterFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
from vidfeat.models.person_bovw_clusters import clusters
HOG = imfeat.HOGLatent(8, 2)
class PersonHandFrameFeature(vidfeat.ClassifierFrameFeature):
#feature = imfeat.PyramidHistogram('lab')
feature = imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(PersonHandFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('person_hand', vidfeat.PersonHandFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
import kernels
class PoorQualityFrameFeature(vidfeat.ClassifierFrameFeature):
feature = imfeat.GradientHistogram()
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(PoorQualityFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
return self.feature(image)
if __name__ == '__main__':
vidfeat._frame_feature_main('poor_quality', vidfeat.PoorQualityFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
class FireFrameFeature(vidfeat.ClassifierFrameFeature):
feature = imfeat.PyramidHistogram('lab')
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(FireFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
return self.feature(image)
if __name__ == '__main__':
vidfeat._frame_feature_main('fire', vidfeat.FireFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
from vidfeat.models.person_bovw_clusters import clusters
HOG = imfeat.HOGLatent(8, 2)
class PersonUpperFrameFeature(vidfeat.ClassifierFrameFeature):
#feature = imfeat.PyramidHistogram('lab')
feature = imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(PersonUpperFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('person_upper', vidfeat.PersonUpperFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
from vidfeat.models.snow_bovw_clusters import clusters
HOG = imfeat.HOGLatent(8, 2)
#import kernels
class SnowFrameFeature(vidfeat.ClassifierFrameFeature):
#feature = imfeat.PyramidHistogram('lab')
feature = imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(SnowFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('snow', vidfeat.SnowFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
import numpy as np
def color_stats(image):
image_hsv = imfeat.convert_image(image, {'mode': 'hsv', 'type': 'numpy', 'dtype': 'float32'})
image_hsv = image_hsv.reshape((image_hsv.shape[0] * image_hsv.shape[1], image_hsv.shape[2]))
image_hsv[:, 0] /= 360 # Rescale
return np.hstack([np.min(image_hsv, 0), np.max(image_hsv, 0), np.mean(image_hsv, 0),
np.median(image_hsv, 0), np.std(image_hsv, 0), imfeat.UniqueColors()(image)])
class GrayFrameFeature(vidfeat.ClassifierFrameFeature):
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(GrayFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
out = color_stats(imfeat.resize_image_max_side(image, 128))
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('gray', vidfeat.GrayFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
class LowLightFrameFeature(vidfeat.ClassifierFrameFeature):
feature = imfeat.MetaFeature(imfeat.GradientHistogram(), imfeat.Histogram('lab', num_bins=4), imfeat.Moments('lab', 2), max_side=128)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(LowLightFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
return self.feature(image)
if __name__ == '__main__':
vidfeat._frame_feature_main('low_light', vidfeat.LowLightFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
class WaterOutdoorFrameFeature(vidfeat.ClassifierFrameFeature):
feature = imfeat.MetaFeature(imfeat.Histogram('lab'), imfeat.GIST(), max_side=128)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(WaterOutdoorFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
return self.feature(image)
if __name__ == '__main__':
vidfeat._frame_feature_main('water_outdoor', vidfeat.WaterOutdoorFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
class VerticalBoxedFrameFeature(vidfeat.ClassifierFrameFeature):
feature = imfeat.BlackBars()
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(VerticalBoxedFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
return self.feature(image)
if __name__ == '__main__':
vidfeat._frame_feature_main('vertical_boxed', vidfeat.VerticalBoxedFrameFeature, remove_bars=True)
from vidfeat.models.synthetic_bovw_clusters import clusters
# import kernels
HOG = imfeat.HOGLatent(8, 2)
class SyntheticFrameFeature(vidfeat.ClassifierFrameFeature):
# feature = imfeat.MetaFeature(imfeat.GradientHistogram(), imfeat.Histogram('lab'))
feature = imfeat.MetaFeature(
imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3),
imfeat.Histogram("lab", num_bins=4),
imfeat.UniqueColors(),
)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight="auto")
self.svm_parameters = [{"C": [10 ** x for x in range(0, 12, 3)]}]
super(SyntheticFrameFeature, self).__init__(classifier=classifier, *args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == "__main__":
vidfeat._frame_feature_main("synthetic", vidfeat.SyntheticFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
from vidfeat.models.synthetic_bovw_clusters import clusters
#import kernels
HOG = imfeat.HOGLatent(8, 2)
class TextFrameFeature(vidfeat.ClassifierFrameFeature):
#feature = imfeat.MetaFeature(imfeat.GradientHistogram(), imfeat.Histogram('lab'))
feature = imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(TextFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('text', vidfeat.TextFrameFeature)
import vidfeat
import imfeat
import sklearn.svm
class HorizontalBoxedFrameFeature(vidfeat.ClassifierFrameFeature):
feature = imfeat.BlackBars()
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(HorizontalBoxedFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def remove_bars(self, image):
pass
def _feature(self, image):
return self.feature(image)
if __name__ == '__main__':
vidfeat._frame_feature_main('horizontal_boxed', vidfeat.HorizontalBoxedFrameFeature, remove_bars=False)
import vidfeat
import imfeat
import sklearn.svm
from vidfeat.models.snow_bovw_clusters import clusters
HOG = imfeat.HOGLatent(8, 2)
class GrassFrameFeature(vidfeat.ClassifierFrameFeature):
#feature = imfeat.PyramidHistogram('lab')
feature = imfeat.BoVW(lambda x: HOG.make_bow_mask(x, clusters), clusters.shape[0], 3)
def __init__(self, *args, **kw):
classifier = sklearn.svm.LinearSVC(class_weight='auto')
self.svm_parameters = [{'C': [10 ** x for x in range(0, 12, 3)]}]
super(GrassFrameFeature, self).__init__(classifier=classifier,
*args, **kw)
def _feature(self, image):
import time
st = time.time()
out = self.feature(imfeat.resize_image_max_side(image, 160))
print time.time() - st
return out
if __name__ == '__main__':
vidfeat._frame_feature_main('grass', vidfeat.GrassFrameFeature)