annotation_append#

Dataset.annotation_append(format_string: str = 'XYWH', category_ids_mapping: dict[int, int] | None = None, label_map: dict[int, str] | None = None) → AnnotationAppender[source]#

Create a context manager to add detection tensors to the current dataset with the AnnotationAppender.append() method, as if the Dataset was a list. After the appending is finished, the appender construct big numpy arrays to concatenate to the dataset’s annotations dataframe.

Note

The dataset object from which this context manager is created is modified inplace, similar to a list append.

Parameters:

format_string – format string for incoming bounding boxes. Depend on your detector conventions. Defaults to “XYWH”.
category_ids_mapping – Optional dictionary to map annotated category ids into the right ids. This is useful for example when a neural network can only use a contiguous label map. Defaults to None
label_map – Optional label map for objects outside the current label map. Must be compatible with the current label map (i.e. no category id clash). Defaults to None.

Returns:

Context manager to easily add detection tensors

Return type:

AnnotationAppender

Example

>>> from lours.utils.doc_utils import dummy_dataset
>>> example = dummy_dataset()
>>> example
Dataset object containing 2 images and 2 objects
Name :
    inside_else_memory
Images root :
    such/serious
Images :
    width  height      relative_path   type  split
id
0     342     136       help/me.jpeg  .jpeg  train
1     377     167  whatever/wait.png   .png  train
Annotations :
    image_id category_str  category_id  ...  box_y_min   box_width  box_height
id                                      ...
0          0         step           15  ...  73.932999   71.552480   42.673983
1          0          why           19  ...   4.567638  248.551257  122.602211

[2 rows x 8 columns]
Label map :
{15: 'step', 19: 'why', 25: 'interview'}
>>> with example.annotation_append(
...     format_string="xyxy", label_map={1: "new_class"}
... ) as appender:
...     appender.append(
...         image_id=0,
...         bbox_coordinates=np.array([[0, 0, 0.5, 0.5]]),
...         category_id=15,
...         confidence=0.5,
...         other_attribute=0,
...     )
...     appender.append(
...         image_id=[1, 0],
...         bbox_coordinates=np.array(
...             [[0.1, 0.1, 0.9, 0.9], [0.2, 0.3, 0.5, 0.5]]
...         ),
...         category_id=np.array([1, 15]),
...         confidence=np.array([0.2, 0.3]),
...         other_attribute=np.array([3, 4]),
...     )
>>> example
Dataset object containing 2 images and 5 objects
Name :
    inside_else_memory
Images root :
    such/serious
Images :
    width  height      relative_path   type  split
id
0     342     136       help/me.jpeg  .jpeg  train
1     377     167  whatever/wait.png   .png  train
Annotations :
    image_id category_str  category_id  ...  box_height  confidence  other_attribute
id                                      ...
0          0         step           15  ...   42.673983         NaN              NaN
1          0          why           19  ...  122.602211         NaN              NaN
2          0         step           15  ...   68.000000         0.5              0.0
3          1    new_class            1  ...  133.600000         0.2              3.0
4          0         step           15  ...   27.200000         0.3              4.0

[5 rows x 10 columns]
Label map :
{1: 'new_class', 15: 'step', 19: 'why', 25: 'interview'}