Skip to content

Latest commit

 

History

History
156 lines (118 loc) · 7.06 KB

object-tracking-3d-ab3dmot.md

File metadata and controls

156 lines (118 loc) · 7.06 KB
Raw

object_tracking_3d_ab3dmot module

The object_tracking_3d_ab3dmot module contains the ObjectTracking3DAb3dmotLearner class, which inherits from the abstract class Learner.

Class ObjectTracking3DAb3dmotLearner

Bases: engine.learners.Learner

The ObjectTracking3DAb3dmotLearner class is an implementation of the AB3DMOT[1] method. Evaluation code is based on the KITTI evaluation development kit[2]. It can be used to perform 3D object tracking based on provided detections.

The ObjectTracking3DAb3dmotLearner class has the following public methods:

ObjectTracking3DAb3dmotLearner constructor

ObjectTracking3DAb3dmotLearner(self, device, max_staleness, min_updates, state_dimensions, measurment_dimensions, state_transition_matrix, measurement_function_matrix, covariance_matrix, process_uncertainty_matrix, iou_threshold)

Constructor parameters:

  • device: {'cpu'}, default='cpu'
    Specifies the device to be used.
  • max_staleness: int, default=2
    Specifies the maximum number of frames when no detections are associated with a track.
  • min_updates: int, default=3
    Specifies the minimal number of updates for tracker to be displayed as output.
  • state_dimensions: int, default=10
    Specifies the number of state dimensions for Kalman filter. Default is 10 for x, y, z, rotation_y, w, l, h, dx, dz, drotation_y.
  • measurement_dimensions: int, default=7
    Specifies the number of measurement dimensions for Kalman filter. Default is 7 for x, y, z, rotation_y, w, l, h.
  • state_transition_matrix: numpy.ndarray, default=None
    Specifies the NumPy state transition matrix for Kalman filter. If None, default one is used.
  • measurement_function_matrix: numpy.ndarray, default=None
    Specifies the NumPy measurement function matrix for Kalman filter. If None, default one is used.
  • covariance_matrix: numpy.ndarray, default=None
    Specifies the NumPy covariance matrix for Kalman filter. If None, default one is used.
  • process_uncertainty_matrix: numpy.ndarray, default=None
    Specifies the NumPy process uncertainity matrix for Kalman filter. If None, default one is used.
  • iou_threshold: float, default=0.01
    Specifies the minimal IoU value to match detection with a tracklet.

ObjectTracking3DAb3dmotLearner.eval

ObjectTracking3DAb3dmotLearner.eval(self, dataset, logging_path, silent, verbose, count)

This method is used to evaluate a trained model on an evaluation dataset. Returns a dictionary containing stats regarding evaluation.

Parameters:

  • dataset: object
    Object that holds the evaluation dataset. Can be of type DatasetIterator.
  • logging_path: str, default=None
    Path to save log files. If set to None, only the console will be used for logging.
  • silent: bool, default=False
    If set to True, disables all printing of evaluation progress reports and other information to STDOUT.
  • verbose: bool, default=False
    If set to True, enables the maximum verbosity.
  • count: int, default=None**
    Specifies the number of sequences to be used for evaluation. If None, the full dataset is used.

ObjectTracking3DAb3dmotLearner.infer

ObjectTracking3DAb3dmotLearner.infer(self, bounding_boxes_3d_list)

This method is used to perform 3D object tracking on a list of 3D bounding boxes predictions. Returns a list of TrackingAnnotation3DList objects if the list of BoundingBox3DList is given or a single TrackingAnnotation3DList if a single BoundingBox3DList is given.

Parameters:

Examples

  • Inference example.

    import os
from opendr.perception.object_tracking_3d import KittiTrackingDatasetIterator
from opendr.perception.object_tracking_3d import ObjectTracking3DAb3dmotLearner

DEVICE = "cpu"
temp_dir = "temp"

dataset = KittiTrackingDatasetIterator(data_path, data_path, "tracking")
learner = ObjectTracking3DAb3dmotLearner()

result = learner.infer(self.dataset[0][0][:5])

print(result)

  • Evaluation example.

    import os
from opendr.perception.object_tracking_3d import KittiTrackingDatasetIterator
from opendr.perception.object_tracking_3d import ObjectTracking3DAb3dmotLearner

DEVICE = "cpu"
temp_dir = "temp"

dataset = KittiTrackingDatasetIterator(data_path, data_path, "tracking")
learner = ObjectTracking3DAb3dmotLearner()

results = learner.eval(dataset, count=2)

for key, val in results.items():
  print(key)
  print(val)

Performance Evaluation

The tests were conducted on the following computational devices:

  • Intel(R) Xeon(R) Gold 6230R CPU on server
  • Nvidia Jetson TX2
  • Nvidia Jetson Xavier AGX
  • Nvidia RTX 2080 Ti GPU on server with Intel Xeon Gold processors

Inference time is measured as the time taken to transfer the input to the model (e.g., from CPU to GPU), run inference using the algorithm, and return results to CPU. Inner FPS refers to the speed of the model when the data is ready. We report FPS (single sample per inference) as the mean of 100 runs.

Full FPS Evaluation of AB3DMOT for classes Car, Pedestrian, Cyclist on KITTI dataset.

Model Object Class TX2 (FPS) Xavier (FPS) RTX 2080 Ti (FPS)
AB3DMOT All 101.26 175.25 344.84

Energy (Joules) of AB3DMOT on embedded devices.

Model Object Class TX2 (Joules) Xavier (Joules)
AB3DMOT All 0.18 0.07

AB3DMOT platform compatibility evaluation.

Platform Test results
x86 - Ubuntu 20.04 (bare installation - CPU) Pass
x86 - Ubuntu 20.04 (bare installation - GPU) Pass
x86 - Ubuntu 20.04 (pip installation) Pass
x86 - Ubuntu 20.04 (CPU docker) Pass
x86 - Ubuntu 20.04 (GPU docker) Pass
NVIDIA Jetson TX2 Pass
NVIDIA Jetson Xavier AGX Pass

References

[1] AB3DMOT: A Baseline for 3D Multi-Object Tracking and New Evaluation Metrics, arXiv. [2] KITTI evaluation development kit.