imsegm.pipelines module

Pipelines for supervised and unsupervised segmentation

Copyright (C) 2014-2018 Jiri Borovec <jiri.borovec@fel.cvut.cz>

imsegm.pipelines.compute_color2d_superpixels_features(image, dict_features, sp_size=30, sp_regul=0.2)

segment image into superpixels and estimate features per superpixel

Parameters

• image (ndarray) – input RGB image

• dict_features (dict(list(str))) – list of features to be extracted

• sp_size (int) – initial size of a superpixel(meaning edge length)

• sp_regul (float) – regularisation in range(0;1) where “0” gives elastic and “1” nearly square segments

Return list(list(int)), [[floats]]

superpixels and related of features

imsegm.pipelines.estim_model_classes_group(list_images, nb_classes, dict_features, sp_size=30, sp_regul=0.2, use_scaler=True, pca_coef=None, model_type='GMM', nb_workers=1)

estimate a model from sequence of input images and return it as result

Parameters

• list_images (list(ndarray)) –

• nb_classes (int) – number of classes

• sp_size (int) – initial size of a superpixel(meaning edge lenght)

• sp_regul (float) – regularisation in range(0;1) where “0” gives elastic and “1” nearly square slic

• dict_features (dict(list(str))) – list of features to be extracted

• pca_coef (float) – range (0, 1) or None

• use_scaler (bool) – whether use a scaler

• model_type (str) – model type

• nb_workers (int) – number of jobs running in parallel

Returns

imsegm.pipelines.pipe_color2d_slic_features_model_graphcut(image, nb_classes, dict_features, sp_size=30, sp_regul=0.2, pca_coef=None, use_scaler=True, estim_model='GMM', gc_regul=1.0, gc_edge_type='model', debug_visual=None)

complete pipe-line for segmentation using superpixels, extracting features and graphCut segmentation

Parameters

• image (ndarray) – input RGB image

• nb_classes (int) – number of classes to be segmented(indexing from 0)

• sp_size (int) – initial size of a superpixel(meaning edge length)

• sp_regul (float) – regularisation in range(0,1) where 0 gives elastic and 1 nearly square slic

• dict_features (dict) – {clr: list(str)}

• pca_coef (float) – range (0, 1) or None

• estim_model (str) – estimating model

• gc_regul (float) – GC regularisation

• gc_edge_type (str) – graphCut edge type

• use_scaler (bool) – using scaler block in pipeline

• l (dict) –

Return list(list(int))

segmentation matrix maping each pixel into a class

>>> np.random.seed(0)
>>> image = np.random.random((125, 150, 3)) / 2.
>>> image[:, :75] += 0.5
>>> segm, seg_soft = pipe_color2d_slic_features_model_graphcut(image, 2, {'color': ['mean']})
>>> segm.shape
(125, 150)
>>> seg_soft.shape
(125, 150, 2)

imsegm.pipelines.pipe_gray3d_slic_features_model_graphcut(image, nb_classes, dict_features, spacing=(12, 1, 1), sp_size=15, sp_regul=0.2, gc_regul=0.1)

complete pipe-line for segmentation using superpixels, extracting features and graphCut segmentation

Parameters

• image (ndarray) – input RGB image

• sp_size (int) – initial size of a superpixel(meaning edge lenght)

• sp_regul (float) – regularisation in range(0;1) where “0” gives elastic and “1” nearly square segments

• nb_classes (int) – number of classes to be segmented(indexing from 0)

• spacing (tuple(int,int,int)) –

• gc_regul (float) – regularisation for GC

Return list(list(int))

segmentation matrix maping each pixel into a class

>>> np.random.seed(0)
>>> image = np.random.random((5, 125, 150)) / 2.
>>> image[:, :, :75] += 0.5
>>> segm = pipe_gray3d_slic_features_model_graphcut(image, 2, {'color': ['mean']})
>>> segm.shape
(5, 125, 150)

imsegm.pipelines.segment_color2d_slic_features_model_graphcut(image, model_pipeline, dict_features, sp_size=30, sp_regul=0.2, gc_regul=1.0, gc_edge_type='model', debug_visual=None)

complete pipe-line for segmentation using superpixels, extracting features and graphCut segmentation

Parameters

• image (ndarray) – input RGB image

• model_pipeline (obj) –

• sp_size (int) – initial size of a superpixel(meaning edge lenght)

• sp_regul (float) – regularisation in range(0;1) where “0” gives elastic and “1” nearly square slic

• dict_features (dict(list(str))) – list of features to be extracted

• gc_regul (float) – GC regularisation

• gc_edge_type (str) – select the GC edge type

• debug_visual – dict

Return list(list(int))

segmentation matrix mapping each pixel into a class

Examples

>>> # UnSupervised:
>>> import imsegm.descriptors as seg_fts
>>> np.random.seed(0)
>>> seg_fts.USE_CYTHON = False
>>> image = np.random.random((125, 150, 3)) / 2.
>>> image[:, :75] += 0.5
>>> model, _ = estim_model_classes_group([image], 2, {'color': ['mean']})
>>> segm, seg_soft = segment_color2d_slic_features_model_graphcut(image, model, {'color': ['mean']})
>>> segm.shape
(125, 150)
>>> seg_soft.shape
(125, 150, 2)

>>> # Supervised:
>>> import imsegm.descriptors as seg_fts
>>> np.random.seed(0)
>>> seg_fts.USE_CYTHON = False
>>> image = np.random.random((125, 150, 3)) / 2.
>>> image[:, 75:] += 0.5
>>> annot = np.zeros(image.shape[:2], dtype=int)
>>> annot[:, 75:] = 1
>>> clf, _, _, _ = train_classif_color2d_slic_features([image], [annot], {'color': ['mean']})
>>> segm, seg_soft = segment_color2d_slic_features_model_graphcut(image, clf, {'color': ['mean']})
>>> segm.shape
(125, 150)
>>> seg_soft.shape
(125, 150, 2)

imsegm.pipelines.train_classif_color2d_slic_features(list_images, list_annots, dict_features, sp_size=30, sp_regul=0.2, clf_name='RandForest', label_purity=0.9, feature_balance='unique', pca_coef=None, nb_classif_search=1, nb_hold_out=2, nb_workers=1)

train classifier on list of annotated images

Parameters

• list_images (list(ndarray)) –

• list_annots (list(ndarray)) –

• sp_size (int) – initial size of a superpixel(meaning edge lenght)

• sp_regul (float) – regularisation in range(0;1) where “0” gives elastic and “1” nearly square segments

• dict_features (dict(list(str))) – list of features to be extracted

• clf_name (str) – selet udsed classifier

• label_purity (float) – set the sample-labels purity for training

• feature_balance (str) – set how to balance datasets

• pca_coef (float) – select PCA coef or None

• nb_classif_search (int) – number of tries for hyper-parameters seach

• nb_hold_out (int) – cross-val leave out

• nb_workers (int) – parallelism

Returns

imsegm.pipelines.wrapper_compute_color2d_slic_features_labels(img_annot, sp_size, sp_regul, dict_features, label_purity)

imsegm.pipelines.CLASSIF_NAME = 'RandForest'

select default Classifier for supervised segmentation

imsegm.pipelines.CLUSTER_METHOD = 'kMeans'

select default Modeling/clustering for unsupervised segmentation

imsegm.pipelines.CROSS_VAL_LEAVE_OUT = 2

define how many images will be left out during cross-validation training

imsegm.pipelines.FTS_SET_SIMPLE = {'color': ('mean', 'std', 'energy')}

select basic features extracted from superpixels

imsegm.pipelines.NB_WORKERS = 1

default number of workers