Heteroskedastic Geospatial Tracking with Distributed Camera Networks

Colin Samplawski, Shiwei Fang, Ziqi Wang, Deepak Ganesan, Mani Srivastava, Benjamin M. Marlin
Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence, PMLR 216:1805-1814, 2023.

Abstract

Visual object tracking has seen significant progress in recent years. However, the vast majority of this work focuses on tracking objects within the image plane of a single camera and ignores the uncertainty associated with predicted object locations. In this work, we focus on the geospatial object tracking problem using data from a distributed camera network. The goal is to predict an object’s track in geospatial coordinates along with uncertainty over the object’s location while respecting communication constraints that prohibit centralizing raw image data. We present a novel single-object geospatial tracking data set that includes high-accuracy ground truth object locations and video data from a network of four cameras. We present a modeling framework for addressing this task including a novel backbone model and explore how uncertainty calibration and fine-tuning through a differentiable tracker affect performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v216-samplawski23a, title = {Heteroskedastic Geospatial Tracking with Distributed Camera Networks}, author = {Samplawski, Colin and Fang, Shiwei and Wang, Ziqi and Ganesan, Deepak and Srivastava, Mani and Marlin, Benjamin M.}, booktitle = {Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence}, pages = {1805--1814}, year = {2023}, editor = {Evans, Robin J. and Shpitser, Ilya}, volume = {216}, series = {Proceedings of Machine Learning Research}, month = {31 Jul--04 Aug}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v216/samplawski23a/samplawski23a.pdf}, url = {https://proceedings.mlr.press/v216/samplawski23a.html}, abstract = {Visual object tracking has seen significant progress in recent years. However, the vast majority of this work focuses on tracking objects within the image plane of a single camera and ignores the uncertainty associated with predicted object locations. In this work, we focus on the geospatial object tracking problem using data from a distributed camera network. The goal is to predict an object’s track in geospatial coordinates along with uncertainty over the object’s location while respecting communication constraints that prohibit centralizing raw image data. We present a novel single-object geospatial tracking data set that includes high-accuracy ground truth object locations and video data from a network of four cameras. We present a modeling framework for addressing this task including a novel backbone model and explore how uncertainty calibration and fine-tuning through a differentiable tracker affect performance.} }
Endnote
%0 Conference Paper %T Heteroskedastic Geospatial Tracking with Distributed Camera Networks %A Colin Samplawski %A Shiwei Fang %A Ziqi Wang %A Deepak Ganesan %A Mani Srivastava %A Benjamin M. Marlin %B Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2023 %E Robin J. Evans %E Ilya Shpitser %F pmlr-v216-samplawski23a %I PMLR %P 1805--1814 %U https://proceedings.mlr.press/v216/samplawski23a.html %V 216 %X Visual object tracking has seen significant progress in recent years. However, the vast majority of this work focuses on tracking objects within the image plane of a single camera and ignores the uncertainty associated with predicted object locations. In this work, we focus on the geospatial object tracking problem using data from a distributed camera network. The goal is to predict an object’s track in geospatial coordinates along with uncertainty over the object’s location while respecting communication constraints that prohibit centralizing raw image data. We present a novel single-object geospatial tracking data set that includes high-accuracy ground truth object locations and video data from a network of four cameras. We present a modeling framework for addressing this task including a novel backbone model and explore how uncertainty calibration and fine-tuning through a differentiable tracker affect performance.
APA
Samplawski, C., Fang, S., Wang, Z., Ganesan, D., Srivastava, M. & Marlin, B.M.. (2023). Heteroskedastic Geospatial Tracking with Distributed Camera Networks. Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 216:1805-1814 Available from https://proceedings.mlr.press/v216/samplawski23a.html.

Related Material