Skip to content

Implementations of 3D metrics for object detection

Notifications You must be signed in to change notification settings

kav-institute/diprolea_3d_metrics

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

11 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Bounding Box Disparity: 3D Metrics for Object Detection With Full Degree of Freedom

Following metrics are part of the implementation:

  • IoU_v: volumetric intersection over union
  • v2v: volume-to-volume distance (shortest distance between the hulls)
  • bbd: bounding box disparity (positive continues combination of IoU and v2v)

  • IoU_p: point-based intersection over union of an underlying pointcloud
  • pd: distance between the centers of the box
  • dd: difference in dimensions
  • od_e: orientation difference using angular difference of euler angles
  • od_R: orientation difference using rotationmatrices

The metrics are available as:

  • stand-alone functions
  • open3d extension
  • ROS-node

Paper

You can find more information in the paper https://arxiv.org/abs/2207.03720

@inproceedings{ ,
	author = {Adam, Michael G. and  Piccolrovazzi, Martin and Eger, Sebastian and Steinbach, Eckehard},
	title = {Bounding Box Disparity: 3D Metrics for Object Detection with Full Degree of Freedom},
	booktitle = {IEEE ICIP 2022},
	year = {2022},
	address = {Bordeaux, France},
	month = {Oct},
	language = {en},
}

Note of the authors (updated):
Googles implementation of IoU (https://github.com/google-research-datasets/Objectron) apparently assumes boxes, which are only one rotation apart. Our implementation does not have this assumption.
Facebooks implementation in PyTorch3D (https://pytorch3d.org/docs/iou3d, Code published end of 2021) seems to be a similar parallel work, however they meshify the boxes and do not state or explain their equations/method (in the corresponding paper IoU isn’t even discussed). We give those explanations in the paper and we think this makes it more easy to follow ;).
Nevertheless, v2v and bbd should be completely new.

Acknowledgement

This work is funded by Germany’s Federal Ministry of Education and Research within the project KIMaps (grant ID #01IS20031C).

About

Implementations of 3D metrics for object detection

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 96.1%
  • Python 3.9%