Skip to content

Commit 38dea74

Browse files
avanettenavanetten
committed
Check for no detections
1 parent 2ffd917 commit 38dea74

File tree

1 file changed

+79
-75
lines changed

1 file changed

+79
-75
lines changed

‎simrdwn/core/simrdwn.py‎

Lines changed: 79 additions & 75 deletions
Original file line numberDiff line numberDiff line change
@@ -1124,7 +1124,7 @@ def run_test(framework='YOLT2',
11241124
# plot
11251125

11261126
# add geo coords to eall boxes?
1127-
if test_add_geo_coords:
1127+
if test_add_geo_coords and len(df_tot) > 0:
11281128
###########################################
11291129
# !!!!! Skip?
11301130
# json = None
@@ -1544,17 +1544,20 @@ def execute(args, train_cmd1, test_cmd_tot, test_cmd_tot2=''):
15441544
rotate_boxes=args.rotate_boxes,
15451545
test_add_geo_coords=args.test_add_geo_coords)
15461546

1547-
# save to csv
1548-
df_tot.to_csv(args.val_df_path_aug, index=False)
1549-
# get number of files
1550-
n_files = len(np.unique(df_tot['Loc_Tmp'].values))
1551-
# n_files = str(len(test_files_locs_list)
1552-
t4 = time.time()
1553-
cmd_time_str = '"Length of time to run test for ' \
1554-
+ str(n_files) + ' files = ' \
1555-
+ str(t4 - t3) + ' seconds\n"'
1556-
print(cmd_time_str[1:-1])
1557-
os.system('echo ' + cmd_time_str + ' >> ' + args.log_file)
1547+
if len(df_tot) == 0:
1548+
print("No detections found!")
1549+
else:
1550+
# save to csv
1551+
df_tot.to_csv(args.val_df_path_aug, index=False)
1552+
# get number of files
1553+
n_files = len(np.unique(df_tot['Loc_Tmp'].values))
1554+
# n_files = str(len(test_files_locs_list)
1555+
t4 = time.time()
1556+
cmd_time_str = '"Length of time to run test for ' \
1557+
+ str(n_files) + ' files = ' \
1558+
+ str(t4 - t3) + ' seconds\n"'
1559+
print(cmd_time_str[1:-1])
1560+
os.system('echo ' + cmd_time_str + ' >> ' + args.log_file)
15581561

15591562
# run again, if desired
15601563
if len(args.weight_file2) > 0:
@@ -1620,69 +1623,70 @@ def execute(args, train_cmd1, test_cmd_tot, test_cmd_tot2=''):
16201623
sliced = False
16211624
print("test data sliced?", sliced)
16221625

1623-
# refine for each plot_thresh
1624-
for plot_thresh_tmp in args.plot_thresh:
1625-
print("Plotting at:", plot_thresh_tmp)
1626-
groupby = 'Image_Path'
1627-
groupby_cat = 'Category'
1628-
df_refine = post_process.refine_df(df_tot,
1629-
groupby=groupby,
1630-
groupby_cat=groupby_cat,
1631-
nms_overlap_thresh=args.nms_overlap_thresh,
1632-
plot_thresh=plot_thresh_tmp,
1633-
verbose=False)
1634-
# make some output plots, if desired
1635-
if len(args.building_csv_file) > 0:
1636-
building_csv_file_tmp = args.building_csv_file.split('.')[0] \
1637-
+ '_plot_thresh_' + str(plot_thresh_tmp).replace('.', 'p') \
1638-
+ '.csv'
1639-
else:
1640-
building_csv_file_tmp = ''
1641-
if args.n_test_output_plots > 0:
1642-
post_process.plot_refined_df(df_refine, groupby=groupby,
1643-
label_map_dict=args.label_map_dict_tot,
1644-
outdir=args.results_dir,
1645-
plot_thresh=plot_thresh_tmp,
1646-
show_labels=bool(
1647-
args.show_labels),
1648-
alpha_scaling=bool(
1649-
args.alpha_scaling),
1650-
plot_line_thickness=args.plot_line_thickness,
1651-
print_iter=5,
1652-
n_plots=args.n_test_output_plots,
1653-
building_csv_file=building_csv_file_tmp,
1654-
shuffle_ims=bool(
1655-
args.shuffle_val_output_plot_ims),
1656-
verbose=False)
1657-
1658-
# geo coords?
1659-
if bool(args.test_add_geo_coords):
1660-
df_refine, json = add_geo_coords.add_geo_coords_to_df(
1661-
df_refine,
1662-
create_geojson=bool(args.save_json),
1663-
inProj_str='epsg:32737', outProj_str='epsg:3857',
1664-
# inProj_str='epsg:4326', outProj_str='epsg:3857',
1665-
verbose=False)
1666-
1667-
# save df_refine
1668-
outpath_tmp = os.path.join(args.results_dir,
1669-
args.val_prediction_df_refine_tot_root_part +
1670-
'_thresh=' + str(plot_thresh_tmp) + '.csv')
1671-
# df_refine.to_csv(args.val_prediction_df_refine_tot)
1672-
df_refine.to_csv(outpath_tmp)
1673-
print("Num objects at thresh:", plot_thresh_tmp, "=",
1674-
len(df_refine))
1675-
# save json
1676-
if bool(args.save_json) and (len(json) > 0):
1677-
output_json_path = os.path.join(args.results_dir,
1678-
args.val_prediction_df_refine_tot_root_part +
1679-
'_thresh=' + str(plot_thresh_tmp) + '.GeoJSON')
1680-
json.to_file(output_json_path, driver="GeoJSON")
1681-
1682-
cmd_time_str = '"Length of time to run refine_test()' + ' ' \
1683-
+ str(time.time() - t8) + ' seconds"'
1684-
print(cmd_time_str[1:-1])
1685-
os.system('echo ' + cmd_time_str + ' >> ' + args.log_file)
1626+
# refine for each plot_thresh (if we have detections)
1627+
if len(df_tot) > 0:
1628+
for plot_thresh_tmp in args.plot_thresh:
1629+
print("Plotting at:", plot_thresh_tmp)
1630+
groupby = 'Image_Path'
1631+
groupby_cat = 'Category'
1632+
df_refine = post_process.refine_df(df_tot,
1633+
groupby=groupby,
1634+
groupby_cat=groupby_cat,
1635+
nms_overlap_thresh=args.nms_overlap_thresh,
1636+
plot_thresh=plot_thresh_tmp,
1637+
verbose=False)
1638+
# make some output plots, if desired
1639+
if len(args.building_csv_file) > 0:
1640+
building_csv_file_tmp = args.building_csv_file.split('.')[0] \
1641+
+ '_plot_thresh_' + str(plot_thresh_tmp).replace('.', 'p') \
1642+
+ '.csv'
1643+
else:
1644+
building_csv_file_tmp = ''
1645+
if args.n_test_output_plots > 0:
1646+
post_process.plot_refined_df(df_refine, groupby=groupby,
1647+
label_map_dict=args.label_map_dict_tot,
1648+
outdir=args.results_dir,
1649+
plot_thresh=plot_thresh_tmp,
1650+
show_labels=bool(
1651+
args.show_labels),
1652+
alpha_scaling=bool(
1653+
args.alpha_scaling),
1654+
plot_line_thickness=args.plot_line_thickness,
1655+
print_iter=5,
1656+
n_plots=args.n_test_output_plots,
1657+
building_csv_file=building_csv_file_tmp,
1658+
shuffle_ims=bool(
1659+
args.shuffle_val_output_plot_ims),
1660+
verbose=False)
1661+
1662+
# geo coords?
1663+
if bool(args.test_add_geo_coords):
1664+
df_refine, json = add_geo_coords.add_geo_coords_to_df(
1665+
df_refine,
1666+
create_geojson=bool(args.save_json),
1667+
inProj_str='epsg:32737', outProj_str='epsg:3857',
1668+
# inProj_str='epsg:4326', outProj_str='epsg:3857',
1669+
verbose=False)
1670+
1671+
# save df_refine
1672+
outpath_tmp = os.path.join(args.results_dir,
1673+
args.val_prediction_df_refine_tot_root_part +
1674+
'_thresh=' + str(plot_thresh_tmp) + '.csv')
1675+
# df_refine.to_csv(args.val_prediction_df_refine_tot)
1676+
df_refine.to_csv(outpath_tmp)
1677+
print("Num objects at thresh:", plot_thresh_tmp, "=",
1678+
len(df_refine))
1679+
# save json
1680+
if bool(args.save_json) and (len(json) > 0):
1681+
output_json_path = os.path.join(args.results_dir,
1682+
args.val_prediction_df_refine_tot_root_part +
1683+
'_thresh=' + str(plot_thresh_tmp) + '.GeoJSON')
1684+
json.to_file(output_json_path, driver="GeoJSON")
1685+
1686+
cmd_time_str = '"Length of time to run refine_test()' + ' ' \
1687+
+ str(time.time() - t8) + ' seconds"'
1688+
print(cmd_time_str[1:-1])
1689+
os.system('echo ' + cmd_time_str + ' >> ' + args.log_file)
16861690

16871691
# remove or zip test_split_dirs to save space
16881692
if len(test_split_dir_list) > 0:

0 commit comments

Comments
 (0)