def dump_word_patches(ibs, vocabdir, invindex, wx_sample, metrics):
"""
Dumps word member patches to disk
"""
wx2_dpath = get_word_dpaths(vocabdir, wx_sample, metrics)
# Write each patch from each annotation to disk
idx2_daid = invindex.idx2_daid
daids = invindex.daids
idx2_dfx = invindex.idx2_dfx
#maws_list = invindex.idx2_wxs[idxs]
# Loop over all annotations skipping the ones without any words in the sample
ax2_idxs = [np.where(idx2_daid == aid_)[0] for aid_ in ut.progiter(daids, 'Building Forward Index: ', freq=100)]
patchdump_iter = ut.progiter(zip(daids, ax2_idxs), freq=1,
lbl='Dumping Selected Patches: ', num=len(daids))
for aid, idxs in patchdump_iter:
wxs_list = invindex.idx2_wxs[idxs]
if len(set(ut.flatten(wxs_list)).intersection(set(wx_sample))) == 0:
# skip this annotation
continue
fx_list = idx2_dfx[idxs]
chip = ibs.get_annot_chips(aid)
chip_kpts = ibs.get_annot_kpts(aid)
nid = ibs.get_annot_name_rowids(aid)
patches, subkpts = ptool.get_warped_patches(chip, chip_kpts)
for fx, wxs, patch in zip(fx_list, wxs_list, patches):
assert len(wxs) == 1, 'did you multiassign the database? If so implement it here too'
for k, wx in enumerate(wxs):
if wx not in wx_sample:
continue
patch_fname = 'patch_nid=%04d_aid=%04d_fx=%04d_k=%d' % (nid, aid, fx, k)
fpath = join(wx2_dpath[wx], patch_fname)
#gtool.imwrite(fpath, patch, fallback=True)
gtool.imwrite_fallback(fpath, patch)
def compute_word_metrics(invindex):
invindex.idx2_wxs = np.array(invindex.idx2_wxs)
wx2_idxs = invindex.wx2_idxs
idx2_dvec = invindex.idx2_dvec
words = invindex.words
wx2_pdist = {}
wx2_wdist = {}
wx2_nMembers = {}
wx2_pdist_stats = {}
wx2_wdist_stats = {}
wordidx_iter = ut.progiter(six.iteritems(wx2_idxs), lbl='Word Dists: ', num=len(wx2_idxs), freq=200)
for _item in wordidx_iter:
wx, idxs = _item
dvecs = idx2_dvec.take(idxs, axis=0)
word = words[wx:wx + 1]
wx2_pdist[wx] = spdist.pdist(dvecs) # pairwise dist between words
wx2_wdist[wx] = ut.euclidean_dist(dvecs, word) # dist to word center
wx2_nMembers[wx] = len(idxs)
for wx, pdist in ut.progiter(six.iteritems(wx2_pdist), lbl='Word pdist Stats: ', num=len(wx2_idxs), freq=2000):
wx2_pdist_stats[wx] = ut.get_stats(pdist)
for wx, wdist in ut.progiter(six.iteritems(wx2_wdist), lbl='Word wdist Stats: ', num=len(wx2_idxs), freq=2000):
wx2_wdist_stats[wx] = ut.get_stats(wdist)
ut.print_stats(wx2_nMembers.values(), 'word members')
metrics = Metrics(wx2_nMembers, wx2_pdist_stats, wx2_wdist_stats)
return metrics
def test_featcfg_combo(ibs, aid, alldictcomb, count, nKpts_list,
cfgstr_list):
for dict_ in ut.progiter(alldictcomb, lbl='FeatCFG Combo: '):
# Set ibs parameters to the current config
for key_, val_ in six.iteritems(dict_):
ibs.cfg.feat_cfg[key_] = val_
cfgstr_ = ibs.cfg.feat_cfg.get_cfgstr()
if count == 0:
# On first run just record info
kpts = ibs.get_annot_kpts(aid)
nKpts_list.append(len(kpts))
cfgstr_list.append(cfgstr_)
if count == 1:
kpts = ibs.get_annot_kpts(aid)
# If second run happens display info
cfgpackstr = utool.packstr(cfgstr_,
textwidth=80,
breakchars=',',
newline_prefix='',
break_words=False,
wordsep=',')
title_suffix = (' len(kpts) = %r \n' % len(kpts)) + cfgpackstr
viz.show_chip(ibs,
aid,
fnum=pt.next_fnum(),
title_suffix=title_suffix,
darken=.8,
ell_linewidth=2,
ell_alpha=.6)
def dump_word_patches(ibs, vocabdir, invindex, wx_sample, metrics):
"""
Dumps word member patches to disk
"""
wx2_dpath = get_word_dpaths(vocabdir, wx_sample, metrics)
# Write each patch from each annotation to disk
idx2_daid = invindex.idx2_daid
daids = invindex.daids
idx2_dfx = invindex.idx2_dfx
#maws_list = invindex.idx2_wxs[idxs]
# Loop over all annotations skipping the ones without any words in the sample
ax2_idxs = [
np.where(idx2_daid == aid_)[0]
for aid_ in ut.progiter(daids, 'Building Forward Index: ', freq=100)
]
patchdump_iter = ut.progiter(zip(daids, ax2_idxs),
freq=1,
lbl='Dumping Selected Patches: ',
num=len(daids))
for aid, idxs in patchdump_iter:
wxs_list = invindex.idx2_wxs[idxs]
if len(set(ut.flatten(wxs_list)).intersection(set(wx_sample))) == 0:
# skip this annotation
continue
fx_list = idx2_dfx[idxs]
chip = ibs.get_annot_chips(aid)
chip_kpts = ibs.get_annot_kpts(aid)
nid = ibs.get_annot_name_rowids(aid)
patches, subkpts = ptool.get_warped_patches(chip, chip_kpts)
for fx, wxs, patch in zip(fx_list, wxs_list, patches):
assert len(
wxs
) == 1, 'did you multiassign the database? If so implement it here too'
for k, wx in enumerate(wxs):
if wx not in wx_sample:
continue
patch_fname = 'patch_nid=%04d_aid=%04d_fx=%04d_k=%d' % (
nid, aid, fx, k)
fpath = join(wx2_dpath[wx], patch_fname)
#gtool.imwrite(fpath, patch, fallback=True)
gtool.imwrite_fallback(fpath, patch)
def get_word_dpaths(vocabdir, wx_sample, metrics):
"""
Gets word folder names and ensure they exist
"""
ut.ensuredir(vocabdir)
wx2_dpath = {wx: join(vocabdir, get_word_dname(wx, metrics)) for wx in wx_sample}
iter_ = ut.progiter(six.itervalues(wx2_dpath), lbl='Ensuring word_dpath: ', freq=200)
for dpath in iter_:
ut.ensuredir(dpath)
return wx2_dpath
def compute_word_metrics(invindex):
invindex.idx2_wxs = np.array(invindex.idx2_wxs)
wx2_idxs = invindex.wx2_idxs
idx2_dvec = invindex.idx2_dvec
words = invindex.words
wx2_pdist = {}
wx2_wdist = {}
wx2_nMembers = {}
wx2_pdist_stats = {}
wx2_wdist_stats = {}
wordidx_iter = ut.progiter(six.iteritems(wx2_idxs),
lbl='Word Dists: ',
num=len(wx2_idxs),
freq=200)
for _item in wordidx_iter:
wx, idxs = _item
dvecs = idx2_dvec.take(idxs, axis=0)
word = words[wx:wx + 1]
wx2_pdist[wx] = spdist.pdist(dvecs) # pairwise dist between words
wx2_wdist[wx] = ut.euclidean_dist(dvecs, word) # dist to word center
wx2_nMembers[wx] = len(idxs)
for wx, pdist in ut.progiter(six.iteritems(wx2_pdist),
lbl='Word pdist Stats: ',
num=len(wx2_idxs),
freq=2000):
wx2_pdist_stats[wx] = ut.get_stats(pdist)
for wx, wdist in ut.progiter(six.iteritems(wx2_wdist),
lbl='Word wdist Stats: ',
num=len(wx2_idxs),
freq=2000):
wx2_wdist_stats[wx] = ut.get_stats(wdist)
ut.print_stats(wx2_nMembers.values(), 'word members')
metrics = Metrics(wx2_nMembers, wx2_pdist_stats, wx2_wdist_stats)
return metrics
def get_word_dpaths(vocabdir, wx_sample, metrics):
"""
Gets word folder names and ensure they exist
"""
ut.ensuredir(vocabdir)
wx2_dpath = {
wx: join(vocabdir, get_word_dname(wx, metrics))
for wx in wx_sample
}
iter_ = ut.progiter(six.itervalues(wx2_dpath),
lbl='Ensuring word_dpath: ',
freq=200)
for dpath in iter_:
ut.ensuredir(dpath)
return wx2_dpath
def test_featcfg_combo(ibs, aid, alldictcomb, count, nKpts_list, cfgstr_list):
for dict_ in ut.progiter(alldictcomb, lbl='FeatCFG Combo: '):
# Set ibs parameters to the current config
for key_, val_ in six.iteritems(dict_):
ibs.cfg.feat_cfg[key_] = val_
cfgstr_ = ibs.cfg.feat_cfg.get_cfgstr()
if count == 0:
# On first run just record info
kpts = ibs.get_annot_kpts(aid)
nKpts_list.append(len(kpts))
cfgstr_list.append(cfgstr_)
if count == 1:
kpts = ibs.get_annot_kpts(aid)
# If second run happens display info
cfgpackstr = utool.packstr(cfgstr_, textwidth=80,
breakchars=',', newline_prefix='',
break_words=False, wordsep=',')
title_suffix = (' len(kpts) = %r \n' % len(kpts)) + cfgpackstr
viz.show_chip(ibs, aid, fnum=pt.next_fnum(),
title_suffix=title_suffix, darken=.8,
ell_linewidth=2, ell_alpha=.6)
def make_wordfigures(ibs, metrics, invindex, figdir, wx_sample, wx2_dpath):
"""
Builds mosaics of patches assigned to words in sample
ouptuts them to disk
"""
from plottool import draw_func2 as df2
import vtool as vt
import parse
vocabdir = join(figdir, 'vocab_patches2')
ut.ensuredir(vocabdir)
dump_word_patches(ibs, vocabdir, invindex, wx_sample, metrics)
# COLLECTING PART --- collects patches in word folders
#vocabdir
seldpath = vocabdir + '_selected'
ut.ensurepath(seldpath)
# stack for show
for wx, dpath in ut.progiter(six.iteritems(wx2_dpath),
lbl='Dumping Word Images:',
num=len(wx2_dpath),
freq=1,
backspace=False):
#df2.rrr()
fpath_list = ut.ls(dpath)
fname_list = [basename(fpath_) for fpath_ in fpath_list]
patch_list = [vt.imread(fpath_) for fpath_ in fpath_list]
# color each patch by nid
nid_list = [
int(parse.parse('{}_nid={nid}_{}', fname)['nid'])
for fname in fname_list
]
nid_set = set(nid_list)
nid_list = np.array(nid_list)
if len(nid_list) == len(nid_set):
# no duplicate names
newpatch_list = patch_list
else:
# duplicate names. do coloring
sortx = nid_list.argsort()
patch_list = np.array(patch_list, dtype=object)[sortx]
fname_list = np.array(fname_list, dtype=object)[sortx]
nid_list = nid_list[sortx]
colors = (255 *
np.array(df2.distinct_colors(len(nid_set)))).astype(
np.int32)
color_dict = dict(zip(nid_set, colors))
wpad, hpad = 3, 3
newshape_list = [
tuple(
(np.array(patch.shape) + (wpad * 2, hpad * 2, 0)).tolist())
for patch in patch_list
]
color_list = [color_dict[nid_] for nid_ in nid_list]
newpatch_list = [
np.zeros(shape) + color[None, None]
for shape, color in zip(newshape_list, color_list)
]
for patch, newpatch in zip(patch_list, newpatch_list):
newpatch[wpad:-wpad, hpad:-hpad, :] = patch
#img_list = patch_list
#bigpatch = vt.stack_image_recurse(patch_list)
#bigpatch = vt.stack_image_list(patch_list, vert=False)
bigpatch = vt.stack_square_images(newpatch_list)
bigpatch_fpath = join(seldpath, basename(dpath) + '_patches.png')
#
def _dictstr(dict_):
str_ = ut.dict_str(dict_, newlines=False)
str_ = str_.replace('\'', '').replace(': ', '=').strip('{},')
return str_
figtitle = '\n'.join([
'wx=%r' % wx,
'stat(pdist): %s' % _dictstr(metrics.wx2_pdist_stats[wx]),
'stat(wdist): %s' % _dictstr(metrics.wx2_wdist_stats[wx]),
])
metrics.wx2_nMembers[wx]
df2.figure(fnum=1, doclf=True, docla=True)
fig, ax = df2.imshow(bigpatch, figtitle=figtitle)
#fig.show()
df2.set_figtitle(figtitle)
df2.adjust_subplots(top=.878, bottom=0)
df2.save_figure(1, bigpatch_fpath)
def make_wordfigures(ibs, metrics, invindex, figdir, wx_sample, wx2_dpath):
"""
Builds mosaics of patches assigned to words in sample
ouptuts them to disk
"""
from plottool import draw_func2 as df2
import vtool as vt
import parse
vocabdir = join(figdir, 'vocab_patches2')
ut.ensuredir(vocabdir)
dump_word_patches(ibs, vocabdir, invindex, wx_sample, metrics)
# COLLECTING PART --- collects patches in word folders
#vocabdir
seldpath = vocabdir + '_selected'
ut.ensurepath(seldpath)
# stack for show
for wx, dpath in ut.progiter(six.iteritems(wx2_dpath), lbl='Dumping Word Images:', num=len(wx2_dpath), freq=1, backspace=False):
#df2.rrr()
fpath_list = ut.ls(dpath)
fname_list = [basename(fpath_) for fpath_ in fpath_list]
patch_list = [gtool.imread(fpath_) for fpath_ in fpath_list]
# color each patch by nid
nid_list = [int(parse.parse('{}_nid={nid}_{}', fname)['nid']) for fname in fname_list]
nid_set = set(nid_list)
nid_list = np.array(nid_list)
if len(nid_list) == len(nid_set):
# no duplicate names
newpatch_list = patch_list
else:
# duplicate names. do coloring
sortx = nid_list.argsort()
patch_list = np.array(patch_list, dtype=object)[sortx]
fname_list = np.array(fname_list, dtype=object)[sortx]
nid_list = nid_list[sortx]
colors = (255 * np.array(df2.distinct_colors(len(nid_set)))).astype(np.int32)
color_dict = dict(zip(nid_set, colors))
wpad, hpad = 3, 3
newshape_list = [tuple((np.array(patch.shape) + (wpad * 2, hpad * 2, 0)).tolist()) for patch in patch_list]
color_list = [color_dict[nid_] for nid_ in nid_list]
newpatch_list = [np.zeros(shape) + color[None, None] for shape, color in zip(newshape_list, color_list)]
for patch, newpatch in zip(patch_list, newpatch_list):
newpatch[wpad:-wpad, hpad:-hpad, :] = patch
#img_list = patch_list
#bigpatch = vt.stack_image_recurse(patch_list)
#bigpatch = vt.stack_image_list(patch_list, vert=False)
bigpatch = vt.stack_square_images(newpatch_list)
bigpatch_fpath = join(seldpath, basename(dpath) + '_patches.png')
#
def _dictstr(dict_):
str_ = ut.dict_str(dict_, newlines=False)
str_ = str_.replace('\'', '').replace(': ', '=').strip('{},')
return str_
figtitle = '\n'.join([
'wx=%r' % wx,
'stat(pdist): %s' % _dictstr(metrics.wx2_pdist_stats[wx]),
'stat(wdist): %s' % _dictstr(metrics.wx2_wdist_stats[wx]),
])
metrics.wx2_nMembers[wx]
df2.figure(fnum=1, doclf=True, docla=True)
fig, ax = df2.imshow(bigpatch, figtitle=figtitle)
#fig.show()
df2.set_figtitle(figtitle)
df2.adjust_subplots(top=.878, bottom=0)
df2.save_figure(1, bigpatch_fpath)
