Data Structures¶

Frame data¶

`StandardFrameData`	Represents a single frame data in intermediate standardized format: Raw frame data -> StandardFrameData -> TransformedFrameData.
`FrameIndexData`	Frame index metadata (for parquet serialization).
`TransformedFrameData`	Represents a single frame data with associated metadata, transformed by Adapters.
`TransformedFrameDataBatch`	Data structure to hold a batch of frame data (PyTorch-friendly).

Trajectories¶

`Trajectory`	Single trajectory data container (numpy) with helpers for processing.
`BatchedTrajectory`	Data container for a batch of trajectories.

Modality containers¶

`CameraData`	Camera sample containing image + calibration matrices.
`LidarData`	Lidar point cloud container used in `StandardFrameData`.
`LidarPointCloud`	Single lidar point cloud after the adapter (numpy-backed).
`BatchedLidarPointCloud`	Batched lidar point clouds in concat-with-batch-index format.
`Detection3D`	Holds 3D detections
`FrameDetections3D`	Holds 3D detections for a single frame
`BatchedFrameDetections3D`	Holds 3D detections for multiple frames

HD map¶

`HDMap`	HD map snapshot in the vehicle frame at a frame's timestamp.
`MapElement`	A single HD map element (polyline, polygon, or point) in vehicle frame.