5268393053

################################################################################
# Copyright (c) 2021 ContinualAI.                                              #
# Copyrights licensed under the MIT License.                                   #
# See the accompanying LICENSE file for terms.                                 #
#                                                                              #
# Date: 19-12-2020                                                             #
# Author(s): ContinualAI                                                       #
# E-mail: contact@continualai.org                                              #
# Website: www.continualai.org                                                 #
################################################################################

""" This module contains the high-level OpenLORIS benchmark/factor generator.
It basically returns a iterable benchmark object :class:`ClassificationScenario`
given a number of configuration parameters."""

from pathlib import Path
from typing import Union, Any, Optional
from typing_extensions import Literal

from avalanche.benchmarks.classic.classic_benchmarks_utils import (
    check_vision_benchmark,
)
from avalanche.benchmarks.datasets.openloris import (
    OpenLORIS as OpenLORISDataset,
)
from avalanche.benchmarks.scenarios.classification_benchmark_creation import (
    create_classification_benchmark_from_filelists,
)
from avalanche.benchmarks.scenarios.classification_scenario import (
    CommonClassificationScenarioType,
)


nbatch = {
    "clutter": 9,
    "illumination": 9,
    "occlusion": 9,
    "pixel": 9,
    "mixture-iros": 12,
}

fac2dirs = {
    "clutter": "batches_filelists/domain/clutter",
    "illumination": "batches_filelists/domain/illumination",
    "occlusion": "batches_filelists/domain/occlusion",
    "pixel": "batches_filelists/domain/pixel",
    "mixture-iros": "batches_filelists/domain/iros",
}


def OpenLORIS(
    *,
    factor: Literal[
        "clutter", "illumination", "occlusion", "pixel", "mixture-iros"
    ] = "clutter",
    train_transform: Optional[Any] = None,
    eval_transform: Optional[Any] = None,
    dataset_root: Optional[Union[str, Path]] = None
):
    """
    Creates a CL benchmark for OpenLORIS.

    If the dataset is not present in the computer, **this method will NOT be
    able automatically download** and store it.

    This generator can be used to obtain scenarios based on different "factors".
    Valid factors include 'clutter', 'illumination', 'occlusion', 'pixel', or
    'mixture-iros'.

    The benchmark instance returned by this method will have two fields,
    `train_stream` and `test_stream`, which can be iterated to obtain
    training and test :class:`Experience`. Each Experience contains the
    `dataset` and the associated task label.

    The task label 0 will be assigned to each experience.

    The benchmark API is quite simple and is uniform across all benchmark
    generators. It is recommended to check the tutorial of the "benchmark" API,
    which contains usage examples ranging from "basic" to "advanced".

    :param factor: OpenLORIS main factors, indicating different environmental
        variations. It can be chosen between 'clutter', 'illumination',
        'occlusion', 'pixel', or 'mixture-iros'. The first three factors are
        included in the ICRA 2020 paper and the last factor (mixture-iros) is
        the benchmark setting for IROS 2019 Lifelong robotic vision competition.
    :param train_transform: The transformation to apply to the training data,
        e.g. a random crop, a normalization or a concatenation of different
        transformations (see torchvision.transform documentation for a
        comprehensive list of possible transformations). Defaults to None.
    :param eval_transform: The transformation to apply to the test data,
        e.g. a random crop, a normalization or a concatenation of different
        transformations (see torchvision.transform documentation for a
        comprehensive list of possible transformations). Defaults to None.
    :param dataset_root: The root path of the dataset.
        Defaults to None, which means that the default location for
        'openloris' will be used.

    :returns: a properly initialized :class:`ClassificationScenario` instance.
    """

    assert factor in nbatch.keys(), (
        "The selected factor is note "
        "recognized: it should be 'clutter',"
        "'illumination', 'occlusion', "
        "'pixel', or 'mixture-iros'."
    )

    # Dataset created just to download it
    dataset = OpenLORISDataset(dataset_root, download=True)

    # Use the root produced by the dataset implementation
    dataset_root = dataset.root

    filelists_bp = fac2dirs[factor] + "/"
    train_failists_paths = []
    for i in range(nbatch[factor]):
        train_failists_paths.append(
            dataset_root
            / filelists_bp
            / ("train_batch_" + str(i).zfill(2) + ".txt")
        )

    factor_obj: CommonClassificationScenarioType = \
        create_classification_benchmark_from_filelists(
            dataset_root,
            train_failists_paths,
            [dataset_root / filelists_bp / "test.txt"],
            task_labels=[0 for _ in range(nbatch[factor])],
            complete_test_set_only=True,
            train_transform=train_transform,
            eval_transform=eval_transform,
        )

    return factor_obj


__all__ = ["OpenLORIS"]

if __name__ == "__main__":
    import sys

    # Untested!
    benchmark_instance = OpenLORIS()
    check_vision_benchmark(benchmark_instance)
    sys.exit(0)

1	################################################################################
2	# Copyright (c) 2021 ContinualAI. #
3	# Copyrights licensed under the MIT License. #
4	# See the accompanying LICENSE file for terms. #
5	# #
6	# Date: 19-12-2020 #
7	# Author(s): ContinualAI #
8	# E-mail: contact@continualai.org #
9	# Website: www.continualai.org #
10	################################################################################
11
12	""" This module contains the high-level OpenLORIS benchmark/factor generator.	4✔
13	It basically returns a iterable benchmark object :class:`ClassificationScenario`
14	given a number of configuration parameters."""
15
16	from pathlib import Path	4✔
17	from typing import Union, Any, Optional	4✔
18	from typing_extensions import Literal	4✔
19
20	from avalanche.benchmarks.classic.classic_benchmarks_utils import (	4✔
21	check_vision_benchmark,
22	)
23	from avalanche.benchmarks.datasets.openloris import (	4✔
24	OpenLORIS as OpenLORISDataset,
25	)
26	from avalanche.benchmarks.scenarios.classification_benchmark_creation import (	4✔
27	create_classification_benchmark_from_filelists,
28	)
29	from avalanche.benchmarks.scenarios.classification_scenario import (	4✔
30	CommonClassificationScenarioType,
31	)
32
33
34	nbatch = {	4✔
35	"clutter": 9,
36	"illumination": 9,
37	"occlusion": 9,
38	"pixel": 9,
39	"mixture-iros": 12,
40	}
41
42	fac2dirs = {	4✔
43	"clutter": "batches_filelists/domain/clutter",
44	"illumination": "batches_filelists/domain/illumination",
45	"occlusion": "batches_filelists/domain/occlusion",
46	"pixel": "batches_filelists/domain/pixel",
47	"mixture-iros": "batches_filelists/domain/iros",
48	}
49
50
51	def OpenLORIS(	4✔
52	*,
53	factor: Literal[
54	"clutter", "illumination", "occlusion", "pixel", "mixture-iros"
55	] = "clutter",
56	train_transform: Optional[Any] = None,
57	eval_transform: Optional[Any] = None,
58	dataset_root: Optional[Union[str, Path]] = None
59	):
60	"""
61	Creates a CL benchmark for OpenLORIS.
62
63	If the dataset is not present in the computer, **this method will NOT be
64	able automatically download** and store it.
65
66	This generator can be used to obtain scenarios based on different "factors".
67	Valid factors include 'clutter', 'illumination', 'occlusion', 'pixel', or
68	'mixture-iros'.
69
70	The benchmark instance returned by this method will have two fields,
71	`train_stream` and `test_stream`, which can be iterated to obtain
72	training and test :class:`Experience`. Each Experience contains the
73	`dataset` and the associated task label.
74
75	The task label 0 will be assigned to each experience.
76
77	The benchmark API is quite simple and is uniform across all benchmark
78	generators. It is recommended to check the tutorial of the "benchmark" API,
79	which contains usage examples ranging from "basic" to "advanced".
80
81	:param factor: OpenLORIS main factors, indicating different environmental
82	variations. It can be chosen between 'clutter', 'illumination',
83	'occlusion', 'pixel', or 'mixture-iros'. The first three factors are
84	included in the ICRA 2020 paper and the last factor (mixture-iros) is
85	the benchmark setting for IROS 2019 Lifelong robotic vision competition.
86	:param train_transform: The transformation to apply to the training data,
87	e.g. a random crop, a normalization or a concatenation of different
88	transformations (see torchvision.transform documentation for a
89	comprehensive list of possible transformations). Defaults to None.
90	:param eval_transform: The transformation to apply to the test data,
91	e.g. a random crop, a normalization or a concatenation of different
92	transformations (see torchvision.transform documentation for a
93	comprehensive list of possible transformations). Defaults to None.
94	:param dataset_root: The root path of the dataset.
95	Defaults to None, which means that the default location for
96	'openloris' will be used.
97
98	:returns: a properly initialized :class:`ClassificationScenario` instance.
99	"""
100
101	assert factor in nbatch.keys(), (	×
102	"The selected factor is note "
103	"recognized: it should be 'clutter',"
104	"'illumination', 'occlusion', "
105	"'pixel', or 'mixture-iros'."
106	)
107
108	# Dataset created just to download it
109	dataset = OpenLORISDataset(dataset_root, download=True)	×
110
111	# Use the root produced by the dataset implementation
112	dataset_root = dataset.root	×
113
114	filelists_bp = fac2dirs[factor] + "/"	×
115	train_failists_paths = []	×
116	for i in range(nbatch[factor]):	×
117	train_failists_paths.append(	×
118	dataset_root
119	/ filelists_bp
120	/ ("train_batch_" + str(i).zfill(2) + ".txt")
121	)
122
NEW 123	factor_obj: CommonClassificationScenarioType = \	×
124	create_classification_benchmark_from_filelists(
125	dataset_root,
126	train_failists_paths,
127	[dataset_root / filelists_bp / "test.txt"],
128	task_labels=[0 for _ in range(nbatch[factor])],
129	complete_test_set_only=True,
130	train_transform=train_transform,
131	eval_transform=eval_transform,
132	)
133
134	return factor_obj	×
135
136
137	__all__ = ["OpenLORIS"]	4✔
138
139	if __name__ == "__main__":	4✔
140	import sys	×
141
142	# Untested!
143	benchmark_instance = OpenLORIS()	×
144	check_vision_benchmark(benchmark_instance)	×
145	sys.exit(0)	×

ContinualAI / avalanche / 5268393053

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous