Module Datasets
This module provides configurations classes and iterators for the lsfb_iso and lsfb_cont dataset. The module is organised in two submodules : One for the continuous dataset and one for the isolated one.
lsfb_cont
This module contains classes helping to manipulate the lsfb_cont dataset. The provided classes are :
- LSFBContConfig : dataclass allowing to configure how the dataset will be loaded
- LSFBContLandmarks : Iterator class loading the lsfb_cont landmarks according to the configuration provided.
- LSFBContLandmarksGenerator : Generator class loading the lsfb_cont landmarks according to the configuration provided.
LSFBContConfig
dataclass
Simple configuration class for the LSFB_CONT dataset. Each instance of the LSFB_CONT dataset is a video with corresponding annotations. Each annotation is made up of a segment (start, end) and a label.
LSFB: French Belgian Sign Language CONT: Continuous videos in LSFB
See LSFBContLandmarks
to load the dataset.
Parameters: |
|
---|
Source code in lsfb-dataset/src/lsfb_dataset/datasets/lsfb_cont/config.py
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 |
|
LSFBContLandmarks
Bases: LSFBContBase
Utility class to load the LSFB CONT Landmarks dataset. The dataset must be already downloaded!
All the landmarks and targets are loaded in memory.
Therefore, iterating over all the instances is fast but consumes a lot of RAM.
If you don't have enough RAM, use the LSFBContLandmarksGenerator
class instead.
Example
my_dataset_config = LSFBContConfig(
root="./my_dataset",
landmarks=['pose', 'left_hand', 'right_hand'],
split="fold_1",
n_labels=750,
segment_level='signs',
segment_unit='frame',
segment_label='sign_gloss',
use_3d=True,
window=(1500, 1200),
)
my_dataset = LSFBContLandmarks(my_dataset_config)
features, target_annotations = dataset[10]
If you did not download the dataset, see lsfb_dataset.Downloader
.
Parameters: |
|
---|
Source code in lsfb-dataset/src/lsfb_dataset/datasets/lsfb_cont/landmarks.py
11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 |
|
LSFBContLandmarksGenerator
Bases: LSFBContBase
Utility class to load the LSFB CONT Landmarks dataset. The dataset must be already downloaded!
All the landmarks and targets are lazily loaded.
Therefore, iterating over all the instances can be a bit slow.
If you have enough RAM and want faster iterations, use the LSFBContLandmarks
class instead.
Example
my_dataset_config = LSFBContConfig(
root="./my_dataset",
landmarks=['pose', 'left_hand', 'right_hand'],
split="fold_1",
n_labels=750,
segment_level='signs',
segment_unit='frame',
segment_label='sign_gloss',
use_3d=True,
window=(1500, 1200),
)
my_dataset = LSFBContLandmarksGenerator(my_dataset_config)
features, target_annotations = dataset[10]
If you did not download the dataset, see lsfb_dataset.Downloader
.
Parameters: |
|
---|
Source code in lsfb-dataset/src/lsfb_dataset/datasets/lsfb_cont/landmarks_generator.py
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 |
|
lsfb_isol
This module contains classes helping to manipulate the lsfb_isol dataset. The provided classes are :
- LSFBIsolConfig : dataclass allowing to configure how the dataset will be loaded
- LSFBIsolLandmarks : Iterator class loading the lsfb_isol landmarks according to the configuration provided.
- LSFBIsolLandmarksGenerator : Generator class loading the lsfb_isol landmarks according to the configuration provided.
LSFBIsolConfig
dataclass
Simple configuration class for the LSFB ISOL Dataset.
LSFB: French Belgian Sign Language ISOL: Isolated signs (videos and poses) in LSFB
If you did not download the dataset, see lsfb_dataset.Downloader
.
Parameters: |
|
---|
Source code in lsfb-dataset/src/lsfb_dataset/datasets/lsfb_isol/config.py
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 |
|
LSFBIsolLandmarks
Bases: LSFBIsolBase
Utility class to load the LSFB ISOL Landmarks dataset. The dataset must be already downloaded!
All the landmarks and targets are loaded in memory.
Therefore, iterating over all the instances is fast but consumes a lot of RAM.
If you don't have enough RAM, use the LSFBIsolLandmarksGenerator
class instead.
Example
my_dataset_config = LSFBIsolConfig(
root="./my_dataset",
split="fold_1",
n_labels=750,
target='sign_gloss',
sequence_max_length=10,
use_3d=True,
)
my_dataset = LSFBIsolLandmarks(my_dataset_config)
features, target = dataset[30]
If you did not download the dataset, see lsfb_dataset.Downloader
.
Parameters: |
|
---|
Source code in lsfb-dataset/src/lsfb_dataset/datasets/lsfb_isol/landmarks.py
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 |
|
LSFBIsolLandmarksGenerator
Bases: LSFBIsolBase
Utility class to load the LSFB ISOL Landmarks dataset. The dataset must be already downloaded!
All the landmarks and targets are lazily loaded.
Therefore, iterating over all the instances can be a bit slow.
If you have enough RAM and want faster iterations, use the LSFBIsolLandmarks
class instead.
Example
my_dataset_config = LSFBIsolConfig(
root="./my_dataset",
split="fold_1",
n_labels=750,
target='sign_gloss',
sequence_max_length=10,
use_3d=True,
)
my_dataset = LSFBIsolLandmarksGenerator(my_dataset_config)
features, target = dataset[30]
Parameters: |
|
---|
Source code in lsfb-dataset/src/lsfb_dataset/datasets/lsfb_isol/landmarks_generator.py
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 |
|