Schema Names
Under the hood, t4-devkit declares an enum called SchemaName.
This enum includes names of each schema table that should be contained in the T4 dataset as .json file.
Note that some schema tables are not mandatory, such as object_ann.json and surface_ann.json.
For these tables, the method called is_optional() returns True and it is OK that these corresponding .json files are not contained in T4 dataset:
SchemaName
An enum to represent schema filenames.
Attributes:
-
ATTRIBUTE–Property of an instance that can change while the category remains the same.
-
CALIBRATED_SENSOR–Definition of a particular sensor as calibrated on a vehicle.
-
CATEGORY–Object categories.
-
EGO_POSE–Ego vehicle pose at at particular timestamp.
-
INSTANCE–An object instance.
-
LOG–Information about the log from which the data aws extracted.
-
MAP–Map data that is stored as binary semantic masks from a top-down view.
-
SAMPLE–A sample is an annotated keyframe at specific Hz.
-
SAMPLE_ANNOTATION–A bounding box defining the position of an object seen in a sample.
-
SAMPLE_DATA–A sensor data e.g. image, pointcloud or radar return.
-
SCENE–A scene is a specific long sequence of consecutive frames extracted from a log.
-
SENSOR–A specific sensor type.
-
VISIBILITY–The visibility of instance is the fraction of annotation visible in all images.
-
LIDARSEG(optional) –The annotation of 3D point cloud segmentation.
-
OBJECT_ANN(optional) –The annotation of a foreground object in an image.
-
SURFACE_ANN(optional) –The annotation of a background object in an image.
-
KEYPOINT(optional) –The annotation of pose keypoints of an object in an image.
-
VEHICLE_STATE(optional) –The annotation of ego vehicle states.