def export_subdatabase_all_annots_new(ibs):
"""
Exports a set with some number of annotations that has good demo examples.
multiple annotations per name and large time variation within names.
Args:
ibs (IBEISController): ibeis controller object
CommandLine:
python -m ibeis.other.ibsfuncs --test-export_subdatabase_all_annots_new --db GIR_Tanya
Example:
>>> # DISABLE_DOCTEST
>>> from ibeis.other.ibsfuncs import * # NOQA
>>> import ibeis
>>> ibs = ibeis.opendb()
>>> print(ibs.get_dbinfo_str())
>>> result = export_subdatabase_all_annots_new(ibs)
>>> # verify results
>>> print(result)
min_num_annots = 500
"""
aid_list = ibs.get_valid_aids()
gid_list = list(set(ibs.get_annot_gids(aid_list)))
from ibeis.dbio import export_subset
def new_nonconflicting_dbpath(ibs):
dpath, dbname = split(ibs.get_dbdir())
base_fmtstr = dbname + '_demo' + 'all_annots' + '_export%d'
new_dbpath = ut.get_nonconflicting_path(base_fmtstr, dpath)
return new_dbpath
#ut.embed()
dbpath = new_nonconflicting_dbpath(ibs)
ibs_dst = ibeis.opendb(dbdir=dbpath, allow_newdir=True)
ibs_src = ibs
export_subset.merge_databases(ibs_src, ibs_dst, gid_list=gid_list)
def export_subdatabase_all_annots_new(ibs):
"""
Exports a set with some number of annotations that has good demo examples.
multiple annotations per name and large time variation within names.
Args:
ibs (IBEISController): ibeis controller object
CommandLine:
python -m ibeis.other.ibsfuncs --test-export_subdatabase_all_annots_new --db GIR_Tanya
Example:
>>> # DISABLE_DOCTEST
>>> from ibeis.other.ibsfuncs import * # NOQA
>>> import ibeis
>>> ibs = ibeis.opendb()
>>> print(ibs.get_dbinfo_str())
>>> result = export_subdatabase_all_annots_new(ibs)
>>> # verify results
>>> print(result)
min_num_annots = 500
"""
aid_list = ibs.get_valid_aids()
gid_list = list(set(ibs.get_annot_gids(aid_list)))
from ibeis.dbio import export_subset
def new_nonconflicting_dbpath(ibs):
dpath, dbname = split(ibs.get_dbdir())
base_fmtstr = dbname + '_demo' + 'all_annots' + '_export%d'
new_dbpath = ut.get_nonconflicting_path(base_fmtstr, dpath)
return new_dbpath
#ut.embed()
dbpath = new_nonconflicting_dbpath(ibs)
ibs_dst = ibeis.opendb(dbdir=dbpath, allow_newdir=True)
ibs_src = ibs
export_subset.merge_databases(ibs_src, ibs_dst, gid_list=gid_list)
def slow_merge_test():
r"""
CommandLine:
python -m ibeis.dbio.export_subset --test-slow_merge_test
Example:
>>> # SLOW_DOCTEST
>>> from ibeis.dbio.export_subset import * # NOQA
>>> result = slow_merge_test()
>>> print(result)
"""
from ibeis.dbio import export_subset
import ibeis
ibs1 = ibeis.opendb('testdb2')
ibs1.fix_invalid_annotmatches()
ibs_dst = ibeis.opendb(
db='testdb_dst2', allow_newdir=True, delete_ibsdir=True)
export_subset.merge_databases(ibs1, ibs_dst)
#ibs_src = ibs1
check_merge(ibs1, ibs_dst)
ibs2 = ibeis.opendb('testdb1')
ibs1.print_dbinfo()
ibs2.print_dbinfo()
ibs_dst.print_dbinfo()
ibs_dst.print_dbinfo()
export_subset.merge_databases(ibs2, ibs_dst)
#ibs_src = ibs2
check_merge(ibs2, ibs_dst)
ibs3 = ibeis.opendb('PZ_MTEST')
export_subset.merge_databases(ibs3, ibs_dst)
#ibs_src = ibs2
check_merge(ibs3, ibs_dst)
ibs_dst.print_dbinfo()
return ibs_dst
def export_testset_for_chuck(ibs, min_num_annots):
"""
Exports a set with some number of annotations that has good demo examples.
multiple annotations per name and large time variation within names.
Args:
ibs (IBEISController): ibeis controller object
CommandLine:
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --dbdir /raid/work2/Turk/PZ_Master --min-num-annots 100
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --dbdir /raid/work2/Turk/PZ_Master --min-num-annots 500
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --dbdir /raid/work2/Turk/GZ_Master --min-num-annots 100
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --dbdir
/raid/work2/Turk/GZ_Master --min-num-annots 500_DOCTEST
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --db GIR_Tanya --min-num-annots 100
Example:
>>> # DISABLE_DOCTEST
>>> from ibeis.other.ibsfuncs import * # NOQA
>>> import ibeis
>>> # build test data
>>> #dbdir = ut.get_argval(('--dbdir',), type_=str, default='testdb1')
>>> min_num_annots = ut.get_argval(('--min-num-annots',), type_=int, default=500)
>>> #ibs = ibeis.opendb('testdb1')
>>> #ibs = ibeis.opendb(dbdir='/raid/work2/Turk/PZ_Master')
>>> ibs = ibeis.opendb() # dbdir=dbdir)
>>> #ibs = ibeis.opendb(dbdir='/raid/work2/Turk/GZ_Master')
>>> print(ibs.get_dbinfo_str())
>>> #ibs = ibeis.opendb('testdb1')
>>> # execute function
>>> result = export_testset_for_chuck(ibs, min_num_annots)
>>> # verify results
>>> print(result)
min_num_annots = 500
"""
import numpy as np
min_num_annots_per_name = 3
max_annot_per_image = 5
#3
#min_num_annots_per_name = 1
#max_annot_per_image = 3000
nid_list = ibs.get_valid_nids()
aids_list = ibs.get_name_aids(nid_list)
nAids_list = list(map(len, aids_list))
nOther_aids_list = ibs.unflat_map(ibs.get_annot_num_contact_aids, aids_list)
invalid_by_num_annots = [num < min_num_annots_per_name for num in nAids_list]
invalid_by_num_others = [any([num > max_annot_per_image for num in nums])
for nums in nOther_aids_list]
invalid_list = ut.or_lists(invalid_by_num_annots, invalid_by_num_others)
valid_nids = ut.filterfalse_items(nid_list, invalid_list)
def get_name_time_variation(ibs, nid_list):
aids_list = ibs.get_name_aids(nid_list)
unixtimes_list = ibs.unflat_map(ibs.get_annot_image_unixtimes, aids_list)
unixtimes_arrs = list(map(np.array, unixtimes_list))
fixtimes_list = [arr[arr > 0] for arr in unixtimes_arrs]
std_list = [np.std(arr) if len(arr) > 1 else 0 for arr in fixtimes_list]
return std_list
std_list = get_name_time_variation(ibs, valid_nids)
sorted_nids = ut.sortedby(valid_nids, std_list, reverse=True)
# Find which names to include
num_annot_cumsum = np.cumsum(ibs.get_name_num_annotations(sorted_nids))
pos_list = np.where(num_annot_cumsum >= min_num_annots)[0]
assert len(pos_list) > 0
nid_list_chosen = sorted_nids[:pos_list[0] + 1]
print('using names:')
print(ibs.get_name_texts(nid_list_chosen))
aids_list_chosen = ibs.get_name_aids(nid_list_chosen)
aid_list_chosen = ut.flatten(aids_list_chosen)
gid_list_chosen = ibs.get_annot_gids(aid_list_chosen)
#ut.debug_duplicate_items(gid_list_chosen)
# make sure not too many other annots are along for the ride
other_aids = ibs.get_annot_contact_aids(aid_list_chosen)
unexpected_aids = list(set(ut.flatten(other_aids)).difference(set(aid_list_chosen)))
print('got %d unexpected_aids' % (len(unexpected_aids),))
from ibeis.dbio import export_subset
def new_nonconflicting_dbpath(ibs):
dpath, dbname = split(ibs.get_dbdir())
base_fmtstr = dbname + '_demo' + str(min_num_annots) + '_export%d'
new_dbpath = ut.get_nonconflicting_path(base_fmtstr, dpath)
return new_dbpath
#ut.embed()
dbpath = new_nonconflicting_dbpath(ibs)
ibs_dst = ibeis.opendb(dbdir=dbpath, allow_newdir=True)
ibs_src = ibs
gid_list = gid_list_chosen
export_subset.merge_databases(ibs_src, ibs_dst, gid_list=gid_list)
DEBUG_NAME = False
if DEBUG_NAME:
ibs.get_name_num_annotations(sorted_nids[0:10])
import plottool as pt
ibeis.viz.viz_name.show_name(ibs, sorted_nids[0])
pt.update()
def export_testset_for_chuck(ibs, min_num_annots):
"""
Exports a set with some number of annotations that has good demo examples.
multiple annotations per name and large time variation within names.
Args:
ibs (IBEISController): ibeis controller object
CommandLine:
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --dbdir /raid/work2/Turk/PZ_Master --min-num-annots 100
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --dbdir /raid/work2/Turk/PZ_Master --min-num-annots 500
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --dbdir /raid/work2/Turk/GZ_Master --min-num-annots 100
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --dbdir
/raid/work2/Turk/GZ_Master --min-num-annots 500_DOCTEST
python -m ibeis.other.ibsfuncs --test-export_testset_for_chuck --db GIR_Tanya --min-num-annots 100
Example:
>>> # DISABLE_DOCTEST
>>> from ibeis.other.ibsfuncs import * # NOQA
>>> import ibeis
>>> # build test data
>>> #dbdir = ut.get_argval(('--dbdir',), type_=str, default='testdb1')
>>> min_num_annots = ut.get_argval(('--min-num-annots',), type_=int, default=500)
>>> #ibs = ibeis.opendb('testdb1')
>>> #ibs = ibeis.opendb(dbdir='/raid/work2/Turk/PZ_Master')
>>> ibs = ibeis.opendb() # dbdir=dbdir)
>>> #ibs = ibeis.opendb(dbdir='/raid/work2/Turk/GZ_Master')
>>> print(ibs.get_dbinfo_str())
>>> #ibs = ibeis.opendb('testdb1')
>>> # execute function
>>> result = export_testset_for_chuck(ibs, min_num_annots)
>>> # verify results
>>> print(result)
min_num_annots = 500
"""
import numpy as np
min_num_annots_per_name = 3
max_annot_per_image = 5
#3
#min_num_annots_per_name = 1
#max_annot_per_image = 3000
nid_list = ibs.get_valid_nids()
aids_list = ibs.get_name_aids(nid_list)
nAids_list = list(map(len, aids_list))
nOther_aids_list = ibs.unflat_map(ibs.get_annot_num_contact_aids,
aids_list)
invalid_by_num_annots = [
num < min_num_annots_per_name for num in nAids_list
]
invalid_by_num_others = [
any([num > max_annot_per_image for num in nums])
for nums in nOther_aids_list
]
invalid_list = ut.or_lists(invalid_by_num_annots, invalid_by_num_others)
valid_nids = ut.filterfalse_items(nid_list, invalid_list)
def get_name_time_variation(ibs, nid_list):
aids_list = ibs.get_name_aids(nid_list)
unixtimes_list = ibs.unflat_map(ibs.get_annot_image_unixtimes,
aids_list)
unixtimes_arrs = list(map(np.array, unixtimes_list))
fixtimes_list = [arr[arr > 0] for arr in unixtimes_arrs]
std_list = [
np.std(arr) if len(arr) > 1 else 0 for arr in fixtimes_list
]
return std_list
std_list = get_name_time_variation(ibs, valid_nids)
sorted_nids = ut.sortedby(valid_nids, std_list, reverse=True)
# Find which names to include
num_annot_cumsum = np.cumsum(ibs.get_name_num_annotations(sorted_nids))
pos_list = np.where(num_annot_cumsum >= min_num_annots)[0]
assert len(pos_list) > 0
nid_list_chosen = sorted_nids[:pos_list[0] + 1]
print('using names:')
print(ibs.get_name_texts(nid_list_chosen))
aids_list_chosen = ibs.get_name_aids(nid_list_chosen)
aid_list_chosen = ut.flatten(aids_list_chosen)
gid_list_chosen = ibs.get_annot_gids(aid_list_chosen)
#ut.debug_duplicate_items(gid_list_chosen)
# make sure not too many other annots are along for the ride
other_aids = ibs.get_annot_contact_aids(aid_list_chosen)
unexpected_aids = list(
set(ut.flatten(other_aids)).difference(set(aid_list_chosen)))
print('got %d unexpected_aids' % (len(unexpected_aids), ))
from ibeis.dbio import export_subset
def new_nonconflicting_dbpath(ibs):
dpath, dbname = split(ibs.get_dbdir())
base_fmtstr = dbname + '_demo' + str(min_num_annots) + '_export%d'
new_dbpath = ut.get_nonconflicting_path(base_fmtstr, dpath)
return new_dbpath
#ut.embed()
dbpath = new_nonconflicting_dbpath(ibs)
ibs_dst = ibeis.opendb(dbdir=dbpath, allow_newdir=True)
ibs_src = ibs
gid_list = gid_list_chosen
export_subset.merge_databases(ibs_src, ibs_dst, gid_list=gid_list)
DEBUG_NAME = False
if DEBUG_NAME:
ibs.get_name_num_annotations(sorted_nids[0:10])
import plottool as pt
ibeis.viz.viz_name.show_name(ibs, sorted_nids[0])
pt.update()
