def compute_statistics_worker(self, samples, grid, pca, gmm):
""" Worker function for computing the sufficient statistics. It takes
each sample, read the points and their locations and computes the
sufficient statistics for the points in each individual bin. The bins
are specified by grid.
"""
for sample in samples:
sample_id = SampID(sample)
# Prepare descriptors: select according to the grid and apply PCA.
siftgeo = read_video_points_from_siftgeo(
self.descriptor_filename % sample)
# TODO Use function get_video_resolution from Dataset
video_infos = get_video_infos(
os.path.join(self.dataset.PREFIX, 'videos',
sample_id.movie + '.avi'))
W, H = video_infos['img_size']
# Filter descriptors into multiple bags according to their location.
bag_xx, bag_ll = self._bin_descriptors(siftgeo, pca, grid, (W, H), (sample_id.bf, sample_id.ef))
all_x = range(1, grid[0] + 1)
all_y = range(1, grid[1] + 1)
all_t = range(1, grid[2] + 1)
for ii, bin in enumerate(product(all_x, all_y, all_t)):
if not self.exist_statistics:
fn = os.path.join(self.temp_path, self.bare_fn % (sample, grid[0], grid[1], grid[2], ii))
try:
with open(fn) as ff:
pass
except IOError:
with open(fn, 'w') as ff:
try:
ss = self.model._compute_statistics(
vstack(bag_xx[bin]), gmm)
except ValueError:
# The current window cell contains no descriptors.
#print 'ValueError %s' % fn
ss = np.array(zeros(self.K + 2 * self.nr_pca_comps * self.K), dtype=np.float32)
ss.tofile(ff)
if self.model.is_spatial_model and not self.exist_spatial_statistics:
# Compute spatial descriptors.
fn = os.path.join(self.temp_path, self.spatial_bare_fn % (sample, grid[0], grid[1], grid[2], ii))
try:
with open(fn) as ff:
pass
except IOError:
with open(fn, 'w') as ff:
try:
ss = self.model._compute_spatial_statistics(vstack(bag_xx[bin]), vstack(bag_ll[bin]), gmm)
except ValueError:
# The current window cell contains no descriptors.
#print 'ValueError %s' % fn
ss = np.array(zeros(self.K + 2 * 3 * self.K), dtype=np.float32)
ss.tofile(ff)
def get_video_dimensions(input_video):
input_video_info = get_video_infos(input_video)
width, height = input_video_info['img_size']
nr_frames = input_video_info['duration'] * input_video_info['fps']
return width, height, nr_frames
def compute_statistics_worker(self, samples, grid, pca, gmm):
""" Worker function for computing the sufficient statistics. It takes
each sample, read the points and their locations and computes the
sufficient statistics for the points in each individual bin. The bins
are specified by grid.
"""
for sample in samples:
sample_id = SampID(sample)
# Prepare descriptors: select according to the grid and apply PCA.
siftgeo = read_video_points_from_siftgeo(self.descriptor_filename %
sample)
# TODO Use function get_video_resolution from Dataset
video_infos = get_video_infos(
os.path.join(self.dataset.PREFIX, 'videos',
sample_id.movie + '.avi'))
W, H = video_infos['img_size']
# Filter descriptors into multiple bags according to their location.
bag_xx, bag_ll = self._bin_descriptors(
siftgeo, pca, grid, (W, H), (sample_id.bf, sample_id.ef))
all_x = range(1, grid[0] + 1)
all_y = range(1, grid[1] + 1)
all_t = range(1, grid[2] + 1)
for ii, bin in enumerate(product(all_x, all_y, all_t)):
if not self.exist_statistics:
fn = os.path.join(
self.temp_path,
self.bare_fn % (sample, grid[0], grid[1], grid[2], ii))
try:
with open(fn) as ff:
pass
except IOError:
with open(fn, 'w') as ff:
try:
ss = self.model._compute_statistics(
vstack(bag_xx[bin]), gmm)
except ValueError:
# The current window cell contains no descriptors.
#print 'ValueError %s' % fn
ss = np.array(
zeros(self.K +
2 * self.nr_pca_comps * self.K),
dtype=np.float32)
ss.tofile(ff)
if self.model.is_spatial_model and not self.exist_spatial_statistics:
# Compute spatial descriptors.
fn = os.path.join(
self.temp_path, self.spatial_bare_fn %
(sample, grid[0], grid[1], grid[2], ii))
try:
with open(fn) as ff:
pass
except IOError:
with open(fn, 'w') as ff:
try:
ss = self.model._compute_spatial_statistics(
vstack(bag_xx[bin]), vstack(bag_ll[bin]),
gmm)
except ValueError:
# The current window cell contains no descriptors.
#print 'ValueError %s' % fn
ss = np.array(zeros(self.K + 2 * 3 * self.K),
dtype=np.float32)
ss.tofile(ff)
def rescale(input_video, max_width, **kwargs):
""" Rescales a video clip by a factor of 2 ** k, such that the width is
smaller than max_width.
Inputs
------
in_video: str
Path to input video.
max_width: int
Maximum width of the rescaled video.
width, height, nr_frames: int, default None
Width, height and number of frames of the original video. If None they
will be computed.
thresh: int, default 10
The maximum difference allowed between the number of frames of the
original movie and the nuber of frames of the rescaled video.
output_video: str, default '/dev/shm/<video_name>'
Path to output video.
get_dimensions: boolean, default False
If True, outputs also the new dimesions of the rescaled video.
Output
------
status: str
The status of the rescale:
'sym_link': the video was small enough; the output is just a
symbolic link.
'rescaled': the video was succesfully rescaled.
'bad_encoding': the input and output video don't have the same
number of frames.
output_video: str
Path to output video.
"""
# Get width, hight, thresh
width = kwargs.get('width', None)
height = kwargs.get('height', None)
nr_frames = kwargs.get('nr_frames', None)
thresh = kwargs.get('thresh', 10)
get_dimensions = kwargs.get('get_dimensions', False)
output_video = kwargs.get(
'output_video', os.path.join('/dev/shm/',
input_video.split('/')[-1]))
try:
if width is None or height is None or nr_frames is None:
input_video_info = get_video_infos(input_video)
width, height = input_video_info['img_size']
nr_frames = input_video_info['duration'] * input_video_info['fps']
except AssertionError, err:
print err
if get_dimensions:
return 'bad_encoding', '', (-1, -1, -1)
else:
return 'bad_encoding', ''
if tt == 1:
# No resizing needed; video is originally small enough.
#os.system('ln -s %s %s' % (input_video, output_video))
output_video = input_video
status = 'sym_link'
else:
# Resize movie clip.
#os.system('ffmpeg -i %s -s %dx%d %s' % (
# input_video, new_width, new_height, output_video))
# Use mencoder to resize videos; it is more reliable. Thanks to
# Danila for the suggestion.
os.system(('mencoder %s -vf scale=%d:%d ' + '-nosound -ovc lavc -o %s')
% # -oac lavc -ac copy
(input_video, new_width, new_height, output_video))
status = 'rescaled'
# Check the time is almost the same.
output_video_info = get_video_infos(output_video)
if abs(output_video_info['duration'] * output_video_info['fps'] -
nr_frames) > thresh:
status = 'bad_encoding'
#print 'Original video: ' + input_video_info.__str__()
#print 'Rescaled video: ' + output_video_info.__str__()
if get_dimensions:
return status, output_video, (new_width, new_height, nr_frames)
else:
return status, output_video
def get_video_dimensions(input_video):
input_video_info = get_video_infos(input_video)
width, height = input_video_info['img_size']
nr_frames = input_video_info['duration'] * input_video_info['fps']
return width, height, nr_frames
def rescale(input_video, max_width, **kwargs):
""" Rescales a video clip by a factor of 2 ** k, such that the width is
smaller than max_width.
Inputs
------
in_video: str
Path to input video.
max_width: int
Maximum width of the rescaled video.
width, height, nr_frames: int, default None
Width, height and number of frames of the original video. If None they
will be computed.
thresh: int, default 10
The maximum difference allowed between the number of frames of the
original movie and the nuber of frames of the rescaled video.
output_video: str, default '/dev/shm/<video_name>'
Path to output video.
get_dimensions: boolean, default False
If True, outputs also the new dimesions of the rescaled video.
Output
------
status: str
The status of the rescale:
'sym_link': the video was small enough; the output is just a
symbolic link.
'rescaled': the video was succesfully rescaled.
'bad_encoding': the input and output video don't have the same
number of frames.
output_video: str
Path to output video.
"""
# Get width, hight, thresh
width = kwargs.get('width', None)
height = kwargs.get('height', None)
nr_frames = kwargs.get('nr_frames', None)
thresh = kwargs.get('thresh', 10)
get_dimensions = kwargs.get('get_dimensions', False)
output_video = kwargs.get('output_video', os.path.join(
'/dev/shm/', input_video.split('/')[-1]))
try:
if width is None or height is None or nr_frames is None:
input_video_info = get_video_infos(input_video)
width, height = input_video_info['img_size']
nr_frames = input_video_info['duration'] * input_video_info['fps']
except AssertionError, err:
print err
if get_dimensions:
return 'bad_encoding', '', (-1, -1, -1)
else:
return 'bad_encoding', ''
if tt == 1:
# No resizing needed; video is originally small enough.
#os.system('ln -s %s %s' % (input_video, output_video))
output_video = input_video
status = 'sym_link'
else:
# Resize movie clip.
#os.system('ffmpeg -i %s -s %dx%d %s' % (
# input_video, new_width, new_height, output_video))
# Use mencoder to resize videos; it is more reliable. Thanks to
# Danila for the suggestion.
os.system(('mencoder %s -vf scale=%d:%d ' +
'-nosound -ovc lavc -o %s') % # -oac lavc -ac copy
(input_video, new_width, new_height, output_video))
status = 'rescaled'
# Check the time is almost the same.
output_video_info = get_video_infos(output_video)
if abs(output_video_info['duration'] * output_video_info['fps']
- nr_frames) > thresh:
status = 'bad_encoding'
#print 'Original video: ' + input_video_info.__str__()
#print 'Rescaled video: ' + output_video_info.__str__()
if get_dimensions:
return status, output_video, (new_width, new_height, nr_frames)
else:
return status, output_video