def main(paths, nb_jobs=NB_THREADS, segm_alpha=MIDDLE_ALPHA_OVERLAP):
logging.info('running...')
assert paths['segms'] != paths['output'], 'overwriting segmentation dir'
assert os.path.basename(paths['images']) != paths['output'], \
'overwriting image dir'
logging.info(tl_expt.string_dict(paths, desc='PATHS'))
if not os.path.exists(paths['output']):
assert os.path.isdir(os.path.dirname(paths['output'])), \
'missing folder: %s' % os.path.dirname(paths['output'])
os.mkdir(paths['output'])
paths_imgs = glob.glob(paths['images'])
logging.info('found %i images in dir "%s"', len(paths_imgs), paths['images'])
_warped_overlap = partial(perform_visu_overlap, paths=paths,
segm_alpha=segm_alpha)
created = []
iterate = tl_expt.WrapExecuteSequence(_warped_overlap, paths_imgs,
nb_jobs=nb_jobs, desc='overlapping')
for r in iterate:
created.append(r)
logging.info('matched and created %i overlaps', np.sum(created))
logging.info('DONE')
def relabel_folder_images(path_images,
path_out,
labels_old,
labels_new,
nb_jobs=1):
""" perform single or multi thread image quantisation
:param [int] labels_old:
:param [int] labels_new:
:param [str] path_images: list of input images
:param path_out: output directory
:param [int] labels_old: list of labels to be replaced
:param [int] labels_new: list of new labels
:param int nb_jobs:
"""
assert os.path.isdir(os.path.dirname(path_images)), \
'missing folder: %s' % path_images
assert os.path.isdir(path_out), 'missing ouput folder: %s' % path_out
path_imgs = sorted(glob.glob(path_images))
logging.info('found %i images', len(path_imgs))
_wrapper_img_relabel = partial(perform_image_relabel,
path_out=path_out,
labels_old=labels_old,
labels_new=labels_new)
iterate = tl_expt.WrapExecuteSequence(_wrapper_img_relabel,
path_imgs,
nb_jobs=nb_jobs,
desc='relabel images')
list(iterate)
def perform_stage(df_group, stage, path_images, path_out):
""" perform cutting images for a particular development stage
and nom them into common image size
:param df_group: input dataframe with ellipse parameters
:param int stage: index of development stage
:param str path_images: path to the image folder
:param str path_out: path to the output folder
"""
logging.info('stage %i listing %i items' % (stage, len(df_group)))
stat_a = NORM_FUNC(df_group['ellipse_a'])
stat_b = NORM_FUNC(df_group['ellipse_b'])
norm_size = (int(stat_b), int(stat_a))
logging.info('normal dimension is %s' % repr(norm_size))
path_out_stage = os.path.join(path_out, str(stage))
if not os.path.isdir(path_out_stage):
os.mkdir(path_out_stage)
wrapper_object = partial(extract_ellipse_object,
path_images=path_images,
path_out=path_out_stage,
norm_size=norm_size)
desc = 'stage %i - size %s' % (stage, norm_size)
iterate = tl_expt.WrapExecuteSequence(wrapper_object,
df_group.iterrows(),
nb_jobs=params['nb_jobs'],
desc=desc)
list(iterate)
def main(params):
""" PIPELINE for matching
:param {str: str} params:
"""
logging.info(tl_expt.string_dict(params, desc='PARAMETERS'))
df_info = pd.read_csv(params['path_infofile'], sep='\t', index_col=0)
df_info = filter_table(df_info, params['path_ellipses'])
logging.info('filtered %i item in table' % len(df_info))
path_csv = os.path.join(params['path_output'], NAME_CSV_RESULTS)
list_evals = []
# get the folder
path_dir_csv = os.path.dirname(params['path_ellipses'])
_wrapper_match = partial(select_optimal_ellipse, path_dir_csv=path_dir_csv)
iterate = tl_expt.WrapExecuteSequence(_wrapper_match,
df_info.iterrows(),
nb_jobs=params['nb_jobs'])
for i, dict_row in enumerate(iterate):
list_evals.append(dict_row)
# every hundreds iteration do export
if i % 100 == 0:
pd.DataFrame(list_evals).to_csv(path_csv)
df_ellipses = pd.DataFrame(list_evals)
df_ellipses.to_csv(path_csv)
def main(params):
""" PIPELINE for rotation
:param {str: str} params:
"""
logging.info('running...')
logging.info(tl_expt.string_dict(params, desc='PARAMETERS'))
list_imgs = sorted(
[p for p in glob.glob(params['path_images']) if os.path.isfile(p)])
logging.info('found images: %i' % len(list_imgs))
if not os.path.isdir(params['path_output']):
os.mkdir(params['path_output'])
wrapper_object = partial(perform_orientation_swap,
path_out=params['path_output'])
dir_name = os.path.dirname(params['path_images'])
iterate = tl_expt.WrapExecuteSequence(wrapper_object,
list_imgs,
nb_jobs=params['nb_jobs'],
desc=dir_name)
list(iterate)
logging.info('DONE')
def main(path_segs, path_out, nb_jobs):
""" the main for creating annotations
:param str path_segs: path with image pattern of images - obj segmentation
:param str path_out:
:param int nb_jobs: number of processes in parallel
:return ndarray:
"""
logging.info('running...')
assert os.path.dirname(path_segs) != path_out, \
'the output dir has to be different then the input object segmentation'
list_imgs = glob.glob(path_segs)
logging.info('found %i images', len(list_imgs))
if not os.path.exists(path_out):
assert os.path.isdir(os.path.dirname(path_out)), \
'missing: %s' % path_out
os.mkdir(path_out)
wrapper_create_annot_centers = partial(create_annot_centers,
path_out_seg=path_out,
path_out_csv=path_out)
iterate = tl_expt.WrapExecuteSequence(wrapper_create_annot_centers,
list_imgs, nb_jobs=nb_jobs,
desc='annotating images')
list(iterate)
logging.info('DONE')
def main(params):
""" PIPELINE candidate clustering
:param {str: any} params:
"""
with open(os.path.join(params['path_expt'],
'config_clustering.json'), 'w') as fp:
json.dump(params, fp)
tl_expt.create_subfolders(params['path_expt'], LIST_SUBDIRS)
list_paths = [params[k] for k in ['path_images', 'path_segms', 'path_centers']]
df_paths = tl_data.find_files_match_names_across_dirs(list_paths)
df_paths.columns = ['path_image', 'path_segm', 'path_points']
df_paths.index = range(1, len(df_paths) + 1)
path_cover = os.path.join(params['path_expt'], run_train.NAME_CSV_TRIPLES)
df_paths.to_csv(path_cover)
logging.info('run clustering...')
df_paths_new = pd.DataFrame()
_wrapper_clustering = partial(cluster_points_draw_export, params=params,
path_out=params['path_expt'])
rows = (dict(row) for idx, row in df_paths.iterrows())
iterate = tl_expt.WrapExecuteSequence(_wrapper_clustering, rows,
nb_jobs=params['nb_jobs'])
for dict_center in iterate:
df_paths_new = df_paths_new.append(dict_center, ignore_index=True)
df_paths_new.set_index('image', inplace=True)
df_paths_new.to_csv(path_cover)
def main(dict_paths, padding=0, use_mask=False, bg_color=None,
nb_jobs=NB_THREADS):
""" the main executable
:param dict_paths:
:param int padding:
:param int nb_jobs:
"""
logging.info('running...')
if not os.path.isdir(dict_paths['output']):
assert os.path.isdir(os.path.dirname(dict_paths['output'])), \
'"%s" should be folder' % dict_paths['output']
logging.debug('creating dir: %s', dict_paths['output'])
os.mkdir(dict_paths['output'])
list_dirs = [dict_paths['annot'], dict_paths['image']]
df_paths = tl_data.find_files_match_names_across_dirs(list_dirs)
logging.info('start cutting images')
wrapper_cutting = partial(export_cut_objects, path_out=dict_paths['output'],
padding=padding, use_mask=use_mask, bg_color=bg_color)
iterate = tl_expt.WrapExecuteSequence(wrapper_cutting,
(row for idx, row in df_paths.iterrows()),
nb_jobs=nb_jobs)
list(iterate)
logging.info('DONE')
def main(paths, nb_jobs=NB_THREADS):
logging.info('running...')
logging.info(tl_expt.string_dict(paths, desc='PATHS'))
if not os.path.exists(paths['output']):
assert os.path.isdir(os.path.dirname(paths['output'])), \
'missing folder: %s' % os.path.dirname(paths['output'])
os.mkdir(paths['output'])
paths_imgs = glob.glob(paths['images'])
logging.info('found %i images in dir "%s"', len(paths_imgs),
paths['images'])
warped_overlap = partial(perform_visu_overlap, paths=paths)
created = []
iterate = tl_expt.WrapExecuteSequence(warped_overlap,
paths_imgs,
nb_jobs=nb_jobs,
desc='overlapping')
for r in iterate:
created.append(r)
logging.info('matched and created %i overlaps', np.sum(created))
logging.info('DONE')
def experiment_loo(params, df_stat, dict_annot, paths_img, path_classif,
path_dump):
imgs_idx_path = list(zip(range(1, len(paths_img) + 1), paths_img))
logging.info('run prediction on training images as Leave-One-Out...')
dict_segms, dict_segms_gc = dict(), dict()
path_out = os.path.join(params['path_exp'], FOLDER_LOO)
path_visu = os.path.join(params['path_exp'], FOLDER_LOO_VISU)
wrapper_segment = partial(retrain_loo_segment_image,
path_classif=path_classif, path_dump=path_dump,
path_out=path_out, path_visu=path_visu)
iterate = tl_expt.WrapExecuteSequence(wrapper_segment, imgs_idx_path,
nb_jobs=params['nb_jobs'],
desc='experiment LOO')
for name, segm, segm_gc in iterate:
dict_segms[name] = segm
dict_segms_gc[name] = segm_gc
gc.collect()
time.sleep(1)
df = eval_segment_with_annot(params, dict_annot, dict_segms, None,
NAME_CSV_SEGM_STAT_RESULT_LOO,
params['nb_jobs'])
df_stat = df_stat.append(get_summary(df, 'segm (LOO)'),
ignore_index=True)
df = eval_segment_with_annot(params, dict_annot, dict_segms_gc, None,
NAME_CSV_SEGM_STAT_RESULT_LOO_GC,
params['nb_jobs'])
df_stat = df_stat.append(get_summary(df, 'segm GC (LOO)'),
ignore_index=True)
path_csv_stat = os.path.join(params['path_exp'], NAME_CSV_SEGM_STAT_RESULTS)
df_stat.set_index(['name']).to_csv(path_csv_stat)
return df_stat
def dataset_load_images_annot_compute_features(params,
show_debug_imgs=SHOW_DEBUG_IMAGES):
""" for all datasets perform the following steps:
1) load image and annotation
2) compute superpixel features and labels
:param {str: ...} params: segmentation parameters
:param bool show_debug_imgs: whether show debug images
:return ({str: ndarray} * 6, [str]):
"""
dict_images, dict_annots = dict(), dict()
dict_slics, dict_features, dict_labels, dict_label_hist = \
dict(), dict(), dict(), dict()
feature_names = list()
# compute features
df_paths = pd.read_csv(params['path_train_list'], index_col=0)
assert all(n in df_paths.columns for n in ['path_image', 'path_annot']), \
'missing required columns in loaded csv file'
wrapper_load_compute = partial(load_image_annot_compute_features_labels,
params=params, show_debug_imgs=show_debug_imgs)
iterate = tl_expt.WrapExecuteSequence(wrapper_load_compute, df_paths.iterrows(),
nb_jobs=params['nb_jobs'],
desc='extract training data')
for name, img, annot, slic, features, labels, label_hist, feature_names in iterate:
dict_images[name] = img
dict_annots[name] = annot
dict_slics[name] = slic
dict_features[name] = features
dict_labels[name] = labels
dict_label_hist[name] = label_hist
# gc.collect(), time.sleep(1)
return dict_images, dict_annots, dict_slics, dict_features, dict_labels, \
dict_label_hist, feature_names
def convert_folder_images(path_images, path_out, path_json=None, nb_jobs=1):
""" perform single or multi thread image quantisation
:param [str] path_images: list of input images
:param path_out: output directory
:param path_json: path to json file
:param int nb_jobs: int
"""
assert os.path.isdir(os.path.dirname(path_images)), \
'input folder does not exist'
path_imgs = sorted(glob.glob(path_images))
logging.info('found %i images', len(path_imgs))
if not os.path.exists(path_out):
assert os.path.isdir(os.path.dirname(path_out)), \
'missing folder: %s' % os.path.dirname(path_out)
os.mkdir(path_out)
dict_colors = load_dict_colours(path_json)
logging.debug('loaded dictionary %s', repr(dict_colors))
wrapper_img_convert = partial(perform_img_convert,
path_out=path_out,
dict_colors=dict_colors)
iterate = tl_expt.WrapExecuteSequence(wrapper_img_convert,
path_imgs,
nb_jobs=nb_jobs,
desc='convert images')
list(iterate)
def experiment_lpo(params,
df_stat,
dict_annot,
idx_paths_img,
path_classif,
path_dump,
nb_holdout,
show_debug_imgs=SHOW_DEBUG_IMAGES):
""" experiment Leave-P-samples-Out
:param {str: ...} params:
:param DF df_stat:
:param {str: ndarray} dict_annot:
:param [str] paths_img:
:param str path_classif:
:param str path_dump:
:param int nb_holdout:
:param bool show_debug_imgs: whether show debug images
:return {}:
"""
logging.info('run prediction on training images as Leave-%i-Out...',
nb_holdout)
dict_segms, dict_segms_gc = dict(), dict()
cv = seg_clf.CrossValidatePOut(len(idx_paths_img), nb_hold_out=nb_holdout)
test_imgs_idx_path = [[idx_paths_img[i] for i in ids] for _, ids in cv]
path_out = os.path.join(params['path_exp'], FOLDER_LPO)
path_visu = os.path.join(params['path_exp'], FOLDER_LPO_VISU) \
if params.get('visual', False) else None
_wrapper_segment = partial(retrain_lpo_segment_image,
path_classif=path_classif,
path_dump=path_dump,
path_out=path_out,
path_visu=path_visu,
show_debug_imgs=show_debug_imgs)
iterate = tl_expt.WrapExecuteSequence(_wrapper_segment,
test_imgs_idx_path,
nb_jobs=params['nb_jobs'],
desc='experiment LPO')
for dict_seg, dict_seg_gc in iterate:
dict_segms.update(dict_seg)
dict_segms_gc.update(dict_seg_gc)
gc.collect()
time.sleep(1)
df = eval_segment_with_annot(params, dict_annot, dict_segms, None,
NAME_CSV_SEGM_STAT_RESULT_LPO % nb_holdout,
params.get('drop_labels',
None), params['nb_jobs'])
df_stat = df_stat.append(get_summary(df, 'segm (L-%i-O)' % nb_holdout),
ignore_index=True)
df = eval_segment_with_annot(params, dict_annot, dict_segms_gc, None,
NAME_CSV_SEGM_STAT_RESULT_LPO_GC % nb_holdout,
params.get('drop_labels',
None), params['nb_jobs'])
df_stat = df_stat.append(get_summary(df, 'segm GC (L-%i-O)' % nb_holdout),
ignore_index=True)
path_csv_stat = os.path.join(params['path_exp'],
NAME_CSV_SEGM_STAT_RESULTS)
df_stat.set_index(['name']).to_csv(path_csv_stat)
return df_stat
def experiment_lpo(params, df_stat, dict_annot, paths_img, path_classif,
path_dump, nb_holdout):
imgs_idx_path = list(zip(range(1, len(paths_img) + 1), paths_img))
logging.info('run prediction on training images as Leave-%i-Out...',
nb_holdout)
dict_segms, dict_segms_gc = dict(), dict()
cv = seg_clf.CrossValidatePOut(len(paths_img), nb_hold_out=nb_holdout)
test_imgs_idx_path = [[imgs_idx_path[i] for i in ids] for _, ids in cv]
path_out = os.path.join(params['path_exp'], FOLDER_LPO)
path_visu = os.path.join(params['path_exp'], FOLDER_LPO_VISU)
wrapper_segment = partial(retrain_lpo_segment_image,
path_classif=path_classif, path_dump=path_dump,
path_out=path_out, path_visu=path_visu)
iterate = tl_expt.WrapExecuteSequence(wrapper_segment, test_imgs_idx_path,
nb_jobs=params['nb_jobs'],
desc='experiment LPO')
for dict_seg, dict_seg_gc in iterate:
dict_segms.update(dict_seg)
dict_segms_gc.update(dict_seg_gc)
gc.collect()
time.sleep(1)
df = eval_segment_with_annot(params, dict_annot, dict_segms, None,
NAME_CSV_SEGM_STAT_RESULT_LPO % nb_holdout,
params['nb_jobs'])
df_stat = df_stat.append(get_summary(df, 'segm (L-%i-O)' % nb_holdout),
ignore_index=True)
df = eval_segment_with_annot(params, dict_annot, dict_segms_gc, None,
NAME_CSV_SEGM_STAT_RESULT_LPO_GC % nb_holdout,
params['nb_jobs'])
df_stat = df_stat.append(get_summary(df, 'segm GC (L-%i-O)' % nb_holdout),
ignore_index=True)
path_csv_stat = os.path.join(params['path_exp'], NAME_CSV_SEGM_STAT_RESULTS)
df_stat.set_index(['name']).to_csv(path_csv_stat)
return df_stat
def main(params):
""" compute the distance among segmented superpixels and given annotation
:param {str: ...} params:
"""
logging.info('running...')
if os.path.isdir(params['path_out']):
logging.info('Missing output dir -> no visual export & results table.')
list_paths = [params['path_images'], params['path_segms']]
df_paths = tl_data.find_files_match_names_across_dirs(list_paths)
df_paths.columns = ['path_image', 'path_segm']
df_dist = pd.DataFrame()
wrapper_eval = partial(compute_boundary_distance, params=params,
path_out=params['path_out'])
iterate = tl_expt.WrapExecuteSequence(wrapper_eval, df_paths.iterrows(),
nb_jobs=params['nb_jobs'],
desc='evaluate SLIC')
for name, dist in iterate:
df_dist = df_dist.append({'name': name, 'mean boundary distance': dist},
ignore_index=True)
df_dist.set_index('name', inplace=True)
if os.path.isdir(params['path_out']):
df_dist.to_csv(os.path.join(params['path_out'], NAME_CSV_DISTANCES))
logging.info('STATISTIC:')
logging.info(df_dist.describe())
logging.info('DONE')
def export_dataset_visual(path_output,
dict_imgs,
dict_segms,
dict_slics,
dict_points,
dict_labels,
nb_jobs=NB_THREADS):
""" visualise complete training dataset by marking labeld points
over image and input segmentation
:param {str: ndarray} dict_imgs:
:param {str: ndarray} dict_segms:
:param {str: ndarray} dict_slics:
:param {str: ndarray} dict_points:
:param {str: ndarray} dict_labels:
:param int nb_jobs: number processing in parallel
"""
logging.info('export training visualisations')
path_out = os.path.join(path_output, FOLDER_POINTS_TRAIN)
gener_args = ((path_out, name, dict_imgs[name], dict_segms[name],
dict_points[name], dict_labels[name], dict_slics[name],
None, '_train') for name in dict_imgs)
iterate = tl_expt.WrapExecuteSequence(wrapper_draw_export_slic_centers,
gener_args,
nb_jobs=nb_jobs,
desc='exporting visualisations')
list(iterate)
def main(params, debug_export=DEBUG_EXPORT):
""" the main entry point
:param {str: ...} params: segmentation parameters
:param bool debug_export: whether export visualisations
"""
logging.getLogger().setLevel(logging.DEBUG)
params = tl_expt.create_experiment_folder(
params, dir_name=NAME_EXPERIMENT, stamp_unique=EACH_UNIQUE_EXPERIMENT)
tl_expt.set_experiment_logger(params['path_exp'])
logging.info(tl_expt.string_dict(params, desc='PARAMETERS'))
# tl_expt.create_subfolders(params['path_exp'], [FOLDER_IMAGE])
df_paths = pd.read_csv(params['path_list'], index_col=0)
logging.info('loaded %i items with columns: %s', len(df_paths),
repr(df_paths.columns.tolist()))
df_paths.dropna(how='any', inplace=True)
# create sub-folders if required
tl_expt.create_subfolders(params['path_exp'], ['input', 'simple'])
dict_segment = create_dict_segmentation(params, None, None, None, None)
tl_expt.create_subfolders(params['path_exp'], [n for n in dict_segment] +
[n + DIR_CENTRE_POSIX for n in dict_segment] +
[n + DIR_VISUAL_POSIX for n in dict_segment])
if debug_export:
list_dirs = [n + DIR_DEBUG_POSIX for n in dict_segment if 'rg2sp' in n]
tl_expt.create_subfolders(params['path_exp'], list_dirs)
_wrapper_segment = partial(image_segmentation, params=params)
iterate = tl_expt.WrapExecuteSequence(_wrapper_segment,
df_paths.iterrows(),
nb_jobs=params['nb_jobs'])
list(iterate)
def quantize_folder_images(path_images,
list_colors=None,
method='color',
px_threshold=THRESHOLD_INVALID_PIXELS,
nb_jobs=1):
""" perform single or multi thread image quantisation
:param str method:
:param float px_threshold:
:param str path_images:, input directory and image pattern for loading
:param list_colors: [(int, int, int)], list of possible colours
:param str method: interpolation method
:param float px_threshold:
:param nb_jobs: int
"""
path_imgs = sorted(glob.glob(path_images))
logging.info('found %i images', len(path_imgs))
if list_colors is None:
dict_colors = see_images_color_info(path_images, px_thr=px_threshold)
list_colors = [c for c in dict_colors]
_wrapper_quantize_img = partial(perform_quantize_image,
method=method,
list_colors=list_colors)
iterate = tl_expt.WrapExecuteSequence(_wrapper_quantize_img,
path_imgs,
nb_jobs=nb_jobs,
desc='quantize images')
list(iterate)
def compute_mean_image(list_img_paths):
iterate = tl_expt.WrapExecuteSequence(tl_data.load_image_2d, list_img_paths,
desc='compute mean image')
imgs = [im[:, :, IMAGE_CHANNEL] for im, _ in iterate]
min_size = np.min([img.shape for img in imgs], axis=0)
imgs = [img[:min_size[0], :min_size[1]] for img in imgs]
img_mean = np.median(imgs, axis=0)
return img_mean
def estim_model_classes_group(list_images,
nb_classes=4,
clr_space='rgb',
sp_size=30,
sp_regul=0.2,
dict_features=FTS_SET_SIMPLE,
pca_coef=None,
proba_type='GMM',
nb_jobs=NB_THREADS):
""" estimate a model from sequence of input images and return it as result
:param [ndarray] list_images:
:param int nb_classes: number of clasees
:param str clr_space: chose the color space
:param int sp_size: initial size of a superpixel(meaning edge lenght)
:param float sp_regul: regularisation in range(0;1) where "0" gives elastic
and "1" nearly square slic
:param {str: [str]} dict_features: list of features to be extracted
:param float pca_coef: range (0, 1) or None
:param str proba_type: model type
:param int nb_jobs: number of jobs running in parallel
:return:
"""
list_slic, list_features = list(), list()
wrapper_compute = partial(compute_color2d_superpixels_features,
sp_size=sp_size,
sp_regul=sp_regul,
dict_features=dict_features,
clr_space=clr_space,
fts_norm=False)
iterate = tl_expt.WrapExecuteSequence(wrapper_compute,
list_images,
nb_jobs=nb_jobs)
for slic, features in iterate:
list_slic.append(slic)
list_features.append(features)
# for img in list_images:
# slic, features = compute_color2d_superpixels_features(img, sp_size,
# sp_regul, dict_features, clr_space, fts_norm=False)
# list_slic.append(slic)
# list_features.append(features)
features = np.concatenate(tuple(list_features), axis=0)
features = np.nan_to_num(features)
# scaling
scaler = preprocessing.StandardScaler()
scaler.fit(features)
features = scaler.transform(features)
pca = None
if pca_coef is not None:
pca = decomposition.PCA(pca_coef)
features = pca.fit_transform(features)
model = seg_gc.estim_class_model(features, nb_classes, proba_type)
return scaler, pca, model
def main_predict(path_classif,
path_pattern_imgs,
path_out,
name='SEGMENT___',
params_local=None):
""" given trained classifier segment new images
:param str path_classif:
:param str path_pattern_imgs:
:param str path_out:
:param str name:
"""
logging.getLogger().setLevel(logging.INFO)
logging.info('running PREDICTION...')
assert path_pattern_imgs is not None
dict_classif = seg_clf.load_classifier(path_classif)
classif = dict_classif['clf_pipeline']
params = dict_classif['params']
if params_local is not None:
params.update({
k: params_local[k]
for k in params_local
if k.startswith('path_') or k.startswith('gc_')
})
path_out, path_visu = prepare_output_dir(path_pattern_imgs,
path_out,
name,
visual=params.get(
'visual', False))
tl_expt.set_experiment_logger(path_out)
logging.info(tl_expt.string_dict(params, desc='PARAMETERS'))
paths_img = sorted(glob.glob(path_pattern_imgs))
logging.info('found %i images on path "%s"', len(paths_img),
path_pattern_imgs)
logging.debug('run prediction...')
show_debug_imgs = params.get('visual', False)
_wrapper_segment = partial(try_segment_image,
params=params,
classif=classif,
path_out=path_out,
path_visu=path_visu,
show_debug_imgs=show_debug_imgs)
list_img_path = list(zip([None] * len(paths_img), paths_img))
iterate = tl_expt.WrapExecuteSequence(_wrapper_segment,
list_img_path,
nb_jobs=params['nb_jobs'],
desc='segmenting images')
for _ in iterate:
gc.collect()
time.sleep(1)
logging.info('prediction DONE')
def main(dict_paths, nb_jobs=NB_THREADS, relabel=True):
""" main evaluation
:param {str: str} dict_paths:
:param int nb_jobs: number of thred running in parallel
:param bool relabel: whether relabel segmentation as sequential
"""
logging.info('running...')
if not os.path.isdir(dict_paths['output']):
assert os.path.isdir(os.path.dirname(dict_paths['output'])), \
'missing folder: %s' % dict_paths['output']
os.mkdir(dict_paths['output'])
name = os.path.basename(os.path.dirname(dict_paths['segm']))
list_dirs = [dict_paths['annot'], dict_paths['segm']]
if dict_paths['image'] != '':
list_dirs.append(dict_paths['image'])
df_paths = tl_data.find_files_match_names_across_dirs(list_dirs)
path_csv = os.path.join(dict_paths['output'], NAME_CVS_PER_IMAGE % name)
df_paths.to_csv(path_csv)
annots, _ = tl_data.load_images_list(df_paths['path_1'].values.tolist())
segms, names = tl_data.load_images_list(df_paths['path_2'].values.tolist())
logging.info('loaded %i annots and %i segms', len(annots), len(segms))
if relabel:
annots = [relabel_sequential(annot)[0] for annot in annots]
segms = list(map(wrapper_relabel_segm, zip(annots, segms)))
path_csv = os.path.join(dict_paths['output'], NAME_CVS_PER_IMAGE % name)
logging.debug('export to "%s"', path_csv)
df_stat = seg_clf.compute_stat_per_image(segms, annots, names, nb_jobs)
df_stat.to_csv(path_csv)
path_csv = os.path.join(dict_paths['output'], NAME_CVS_OVERALL % name)
logging.debug('export to "%s"', path_csv)
df_desc = df_stat.describe()
logging.info(df_desc.T[['count', 'mean', 'std']])
df_desc.to_csv(path_csv)
path_visu = os.path.join(dict_paths['output'], '%s__visual' % name)
if not os.path.isdir(path_visu):
os.mkdir(path_visu)
# for idx, row in df_paths.iterrows():
# export_visual(row, path_visu)
wrapper_visual = partial(export_visual, path_out=path_visu)
iterate = tl_expt.WrapExecuteSequence(
wrapper_visual, (row for idx, row in df_paths.iterrows()),
nb_jobs=nb_jobs)
list(iterate)
logging.info('DONE')
def evaluate_detection_stage(df_paths, stage, path_info, path_out, nb_jobs=1):
""" evaluate center detection for particular list of stages
:param df_paths:
:param [int] stage:
:param str path_info:
:param str path_out:
:param int nb_jobs:
:return DF:
"""
logging.info('evaluate stages: %s', repr(stage))
str_stage = '-'.join(map(str, stage))
path_csv = os.path.join(path_out, NAME_CSV_ANNOT_STAGE % str_stage)
if not os.path.exists(path_csv) or FORCE_RELOAD:
df_slices_info = seg_annot.load_info_group_by_slices(path_info, stage)
logging.debug('export slices_info to "%s"', path_csv)
df_slices_info.to_csv(path_csv)
else:
logging.debug('loading slices_info from "%s"', path_csv)
df_slices_info = pd.read_csv(path_csv, index_col=0)
if len(df_slices_info) == 0:
return df_paths
# df_paths = pd.merge(df_paths, df_slices_info, how='inner',
# left_index=True, right_index=True)
df_eval = pd.DataFrame()
path_annot = os.path.join(path_out, FOLDER_ANNOT % str_stage)
path_visu = os.path.join(path_out, FOLDER_ANNOT_VISUAL % str_stage)
list_dirs = [os.path.basename(p) for p in [path_annot, path_visu]]
logging.debug('create sub-dirs: %s', repr(list_dirs))
tl_expt.create_subfolders(path_out, list_dirs)
# perfom on new images
stage_prefix = '[stage-%s] ' % str_stage
logging.info('start section %s - load_center_evaluate ...', stage_prefix)
wrapper_detection = partial(load_center_evaluate,
df_annot=df_slices_info,
path_annot=path_annot,
path_visu=path_visu,
col_prefix=stage_prefix)
iterate = tl_expt.WrapExecuteSequence(wrapper_detection,
df_paths.iterrows(),
nb_jobs=nb_jobs)
for dict_eval in iterate:
df_eval = df_eval.append(dict_eval, ignore_index=True)
df_eval.to_csv(os.path.join(path_out, NAME_CSV_TRIPLES_TEMP))
# gc.collect(), time.sleep(1)
return df_eval
def estim_model_classes_group(list_images,
nb_classes,
dict_features,
sp_size=30,
sp_regul=0.2,
b_scaler=True,
pca_coef=None,
model_type='GMM',
nb_jobs=NB_THREADS):
""" estimate a model from sequence of input images and return it as result
:param [ndarray] list_images:
:param int nb_classes: number of classes
:param int sp_size: initial size of a superpixel(meaning edge lenght)
:param float sp_regul: regularisation in range(0;1) where "0" gives elastic
and "1" nearly square slic
:param {str: [str]} dict_features: list of features to be extracted
:param float pca_coef: range (0, 1) or None
:param bool b_scaler: whether use a scaler
:param str model_type: model type
:param int nb_jobs: number of jobs running in parallel
:return:
"""
list_slic, list_features = list(), list()
_wrapper_compute = partial(compute_color2d_superpixels_features,
sp_size=sp_size,
sp_regul=sp_regul,
dict_features=dict_features,
fts_norm=False)
iterate = tl_expt.WrapExecuteSequence(_wrapper_compute,
list_images,
desc='compute SLIC & features',
nb_jobs=nb_jobs)
for slic, features in iterate:
list_slic.append(slic)
list_features.append(features)
# for img in list_images:
# slic, features = compute_color2d_superpixels_features(img, sp_size,
# sp_regul, dict_features, fts_norm=False)
# list_slic.append(slic)
# list_features.append(features)
features = np.concatenate(tuple(list_features), axis=0)
features = np.nan_to_num(features)
model = seg_gc.estim_class_model(features, nb_classes, model_type,
pca_coef, b_scaler)
return model, list_features
def experiment_single_gmm(params, paths_img, path_out, path_visu):
imgs_idx_path = list(zip([None] * len(paths_img), paths_img))
logging.info('Perform image segmentation as single image in each time')
dict_segms_gmm = {}
wrapper_segment = partial(segment_image_independent,
params=params,
path_out=path_out,
path_visu=path_visu)
iterate = tl_expt.WrapExecuteSequence(wrapper_segment,
imgs_idx_path,
nb_jobs=params['nb_jobs'],
desc='experiment single GMM')
for name, segm in iterate:
dict_segms_gmm[name] = segm
return dict_segms_gmm
def perform_predictions(params, paths_img, classif):
logging.info('run prediction on training images...')
imgs_idx_path = list(zip(range(1, len(paths_img) + 1), paths_img))
dict_segms, dict_segms_gc = dict(), dict()
path_out = os.path.join(params['path_exp'], FOLDER_SEGM)
path_visu = os.path.join(params['path_exp'], FOLDER_SEGM_VISU)
wrapper_segment = partial(segment_image, params=params, classif=classif,
path_out=path_out, path_visu=path_visu)
iterate = tl_expt.WrapExecuteSequence(wrapper_segment, imgs_idx_path,
nb_jobs=params['nb_jobs'],
desc='image segm: prediction')
for name, segm, segm_gc in iterate:
dict_segms[name] = segm
dict_segms_gc[name] = segm_gc
return dict_segms, dict_segms_gc
def main(params):
""" PIPELINE for new detections
:param {str: str} paths:
"""
logging.info('running...')
params = run_train.prepare_experiment_folder(params, FOLDER_EXPERIMENT)
# run_train.check_pathes_patterns(paths)
tl_expt.set_experiment_logger(params['path_expt'])
logging.info('COMPUTER: \n%s', repr(os.uname()))
logging.info(tl_expt.string_dict(params, desc='PARAMETERS'))
tl_expt.create_subfolders(params['path_expt'], LIST_SUBFOLDER)
path_csv = os.path.join(params['path_expt'], NAME_CSV_TRIPLES)
df_paths = get_csv_triplets(params['path_list'],
path_csv,
params['path_images'],
params['path_segms'],
force_reload=FORCE_RERUN)
dict_classif = seg_clf.load_classifier(params['path_classif'])
params_clf = dict_classif['params']
params_clf.update(params)
logging.info(tl_expt.string_dict(params, desc='UPDATED PARAMETERS'))
# perform on new images
df_stat = pd.DataFrame()
wrapper_detection = partial(load_compute_detect_centers,
params=params_clf,
path_classif=params['path_classif'],
path_output=params['path_expt'])
iterate = tl_expt.WrapExecuteSequence(wrapper_detection,
df_paths.iterrows(),
nb_jobs=params['nb_jobs'])
for dict_center in iterate:
df_stat = df_stat.append(dict_center, ignore_index=True)
df_stat.to_csv(os.path.join(params['path_expt'],
NAME_CSV_TRIPLES_TEMP))
df_stat.set_index(['image'], inplace=True)
df_stat.to_csv(os.path.join(params['path_expt'], NAME_CSV_TRIPLES))
logging.info('STATISTIC: \n %s', repr(df_stat.describe()))
logging.info('DONE')
def quantize_folder_images(path_images, label, nb_jobs=1):
""" perform single or multi thread image quantisation
:param [str] path_images: list of image paths
:param int nb_jobs:
"""
assert os.path.isdir(os.path.dirname(path_images)), \
'input folder does not exist: %s' % os.path.dirname(path_images)
path_imgs = sorted(glob.glob(path_images))
logging.info('found %i images', len(path_imgs))
_wrapper_img_inpaint = partial(perform_img_inpaint, labels=label)
iterate = tl_expt.WrapExecuteSequence(_wrapper_img_inpaint,
path_imgs,
nb_jobs=nb_jobs,
desc='quantise images')
list(iterate)
def experiment_group_gmm(params,
paths_img,
path_out,
path_visu,
show_debug_imgs=SHOW_DEBUG_IMAGES):
logging.info('load all images')
list_images = [
load_image(path_img, params['img_type']) for path_img in paths_img
]
imgs_idx_path = list(zip([None] * len(paths_img), paths_img))
logging.info('Estimate image segmentation from whole sequence of images')
params['path_model'] = os.path.join(params['path_exp'], NAME_DUMP_MODEL)
if os.path.isfile(params['path_model']) and not FORCE_RECOMP_DATA:
model, _, _ = load_model(params['path_model'])
else:
model, _ = seg_pipe.estim_model_classes_group(
list_images,
nb_classes=params['nb_classes'],
dict_features=params['features'],
sp_size=params['slic_size'],
sp_regul=params['slic_regul'],
pca_coef=params['pca_coef'],
model_type=params['estim_model'])
save_model(params['path_model'], model)
logging.info('Perform image segmentation from group model')
_wrapper_segment = partial(segment_image_model,
params=params,
model=model,
path_out=path_out,
path_visu=path_visu,
show_debug_imgs=show_debug_imgs)
iterate = tl_expt.WrapExecuteSequence(_wrapper_segment,
imgs_idx_path,
nb_jobs=params['nb_jobs'],
desc='experiment group GMM')
# dict_segms_group = {}
# for name, segm in iterate:
# dict_segms_group[name] = segm
dict_segms_group = dict(iterate)
gc.collect()
time.sleep(1)
return dict_segms_group
def main(params):
df_paths = tl_data.find_files_match_names_across_dirs([params['path_images'],
params['path_segms'],
params['path_centers']])
df_paths.columns = ['path_image', 'path_segm', 'path_centers']
df_paths.index = range(1, len(df_paths) + 1)
if not os.path.exists(params['path_output']):
assert os.path.exists(os.path.dirname(params['path_output'])), \
'missing folder: "%s"' % os.path.dirname(params['path_output'])
os.mkdir(params['path_output'])
df_slices_info = seg_annot.load_info_group_by_slices(params['path_infofile'],
params['stages'])
_wrapper_export = partial(export_figure, df_slices_info=df_slices_info,
path_out=params['path_output'])
iterate = tl_expt.WrapExecuteSequence(_wrapper_export, df_paths.iterrows(),
nb_jobs=params['nb_jobs'])
list(iterate)