fiftyone.utils.kitti

module documentation

(source)

Utilities for working with datasets in KITTI format.

Class	`KITTIAnnotationWriter`	Class for writing annotations in KITTI detection format.
Class	`KITTIDetectionDatasetExporter`	Exporter that writes KITTI detection datasets to disk.
Class	`KITTIDetectionDatasetImporter`	Importer for KITTI detection datasets stored on disk.
Function	`download_kitti_detection_dataset`	Downloads the KITTI object detection dataset from the web.
Function	`download_kitti_multiview_dataset`	Downloads and prepares the multiview KITTI dataset.
Function	`load_kitti_detection_annotations`	Loads the KITTI detection annotations from the given TXT file.
Variable	`logger`	Undocumented
Variable	`o3d`	Undocumented
Function	`_convert_velodyne_to_pcd`	Undocumented
Function	`_do_conversion`	Undocumented
Function	`_download_and_unpack_kitti_zip`	Undocumented
Function	`_load_calibration_matrices`	Undocumented
Function	`_load_kitti_annotations`	Undocumented
Function	`_make_kitti_detection_row`	Undocumented
Function	`_normalize_3d_detections`	Undocumented
Function	`_parse_kitti_detection_row`	Undocumented
Function	`_prepare_kitti_split`	Undocumented
Function	`_proj_3d_to_right_camera`	Undocumented
Function	`_roty`	Undocumented
Function	`_swap_coordinates`	Undocumented
Function	`_write_fo3d_files`	Undocumented
Constant	`_CALIB`	Undocumented
Constant	`_LABELS`	Undocumented
Constant	`_LEFT_IMAGES`	Undocumented
Constant	`_RIGHT_IMAGES`	Undocumented
Constant	`_VELODYNE`	Undocumented

def download_kitti_detection_dataset(dataset_dir, splits=None, scratch_dir=None, overwrite=False, cleanup=False): (source) ¶

Downloads the KITTI object detection dataset from the web.

The dataset will be organized on disk in as follows:

dataset_dir/
    train/
        data/
            000000.png
            000001.png
            ...
        labels/
            000000.txt
            000001.txt
            ...
    test/
        data/
            000000.png
            000001.png
            ...

Parameters
dataset_dir	the directory in which to construct the dataset
splits:`None`	the split or list of splits to download. Supported values are `("train", "test")`
scratch_dir:`None`	a scratch directory to use to download any necessary temporary files
overwrite:`False`	whether to redownload the zips if they already exist
cleanup:`False`	whether to delete the downloaded zips and scratch directory

def download_kitti_multiview_dataset(dataset_dir, splits=None, scratch_dir=None, overwrite=False, cleanup=False, num_workers=None): (source) ¶

Downloads and prepares the multiview KITTI dataset.

The dataset will be organized on disk in as follows, with each split stored in :ref:`FiftyOneDataset format <FiftyOneDataset-import>`:

dataset_dir/
    train/
        labels/
            000000.txt
            000001.txt
            ...
        calib/
            000000.txt
            000001.txt
            ...
        left/
            000000.png
            000001.png
            ...
        right/
            000000.png
            000001.png
            ...
        velodyne/
            000000.bin
            000001.bin
            ...
        pcd/
            000000.pcd
            000001.pcd
            ...
        metadata.json
        samples.json
    test/
        ...

Parameters
dataset_dir	the directory in which to construct the dataset
splits:`None`	the split or list of splits to download. Supported values are `("train", "test")`
scratch_dir:`None`	a scratch directory to use to download any necessary temporary files
overwrite:`False`	whether to redownload/regenerate files if they already exist
cleanup:`False`	whether to delete the downloaded zips and scratch directory
num_workers:`None`	a suggested number of processes to use when converting LiDAR to PCD

def load_kitti_detection_annotations(txt_path, frame_size, extra_attrs=True): (source) ¶

Loads the KITTI detection annotations from the given TXT file.

See :ref:`this page <KITTIDetectionDataset-import>` for format details.

Parameters
txt_path	the path to the annotations TXT file
frame_size	the `(width, height)` of the image
extra_attrs:`True`	whether to load extra annotation attributes onto the imported labels. Supported values are: `True`: load all extra attributes found `False`: do not load extra attributes a name or list of names of specific attributes to load
Returns
a `fiftyone.core.detections.Detections` instance

logger = (source) ¶

Undocumented

o3d = (source) ¶

Undocumented

def _convert_velodyne_to_pcd(velodyne_map, calib_map, pcd_dir, uuids, overwrite=False, num_workers=None): (source) ¶

Undocumented

def _do_conversion(input): (source) ¶

Undocumented

def _download_and_unpack_kitti_zip(data, dataset_dir, scratch_dir, splits, name, cleanup=False, overwrite=False): (source) ¶

Undocumented

def _load_calibration_matrices(inpath): (source) ¶

Undocumented

def _load_kitti_annotations(labels_path, frame_size): (source) ¶

Undocumented

def _make_kitti_detection_row(detection, frame_size): (source) ¶

Undocumented

def _normalize_3d_detections(detections): (source) ¶

Undocumented

def _parse_kitti_detection_row(row, frame_size, extra_attrs): (source) ¶

Undocumented

def _prepare_kitti_split(split_dir, overwrite=False, num_workers=None): (source) ¶

Undocumented

def _proj_3d_to_right_camera(detections3d, calib, frame_size): (source) ¶

Undocumented

def _roty(t): (source) ¶

Undocumented

def _swap_coordinates(vec): (source) ¶

Undocumented

def _write_fo3d_files(pcd_dir, fo3d_dir, overwrite=False, abs_paths=False): (source) ¶

Undocumented

_CALIB: dict = (source) ¶

Undocumented

Value

{'url': 'https://s3.eu-central-1.amazonaws.com/avg-kitti/data_object_calib.zip',
 'contents': {'train': ['training', 'calib'], 'test': ['testing', 'calib']}}

_LABELS: dict = (source) ¶

Undocumented

Value

{'url': 'https://s3.eu-central-1.amazonaws.com/avg-kitti/data_object_label_2.zip'↵
,
 'contents': {'train': ['training', 'label_2']}}

_LEFT_IMAGES: dict = (source) ¶

Undocumented

Value

{'url': 'https://s3.eu-central-1.amazonaws.com/avg-kitti/data_object_image_2.zip'↵
,
 'contents': {'train': ['training', 'image_2'], 'test': ['testing', 'image_2']}}

_RIGHT_IMAGES: dict = (source) ¶

Undocumented

Value

{'url': 'https://s3.eu-central-1.amazonaws.com/avg-kitti/data_object_image_3.zip'↵
,
 'contents': {'train': ['training', 'image_3'], 'test': ['testing', 'image_3']}}

_VELODYNE: dict = (source) ¶

Undocumented

Value

{'url': 'https://s3.eu-central-1.amazonaws.com/avg-kitti/data_object_velodyne.zi↵
p',
 'contents': {'train': ['training', 'velodyne'],
              'test': ['testing', 'velodyne']}}