Severity of Catastrophic Forgetting in Object Detection for Autonomous Driving

Christian Witte, Christian Witte, René Schuster, Syed Bukhari, Patrick Trampert, Didier Stricker, Didier Stricker, Georg Schneider

2023

Abstract

Incorporating unseen data in pre-trained neural networks remains a challenging endeavor, as complete retraining is often impracticable. Yet, training the networks sequentially on data with different distributions can lead to performance degradation for previously learned data, known as catastrophic forgetting. The sequential training paradigm and the mitigation of catastrophic forgetting are subject to Continual Learning (CL). The phenomenon of forgetting poses a challenge for applications with changing distributions and prediction objectives, including Autonomous Driving (AD). Our work aims to illustrate the severity of catastrophic forgetting for object detection for class- and domain-incremental learning. We propose four hypotheses, as we investigate the impact of the ordering of sequential increments and the underlying data distribution of AD datasets. Further, the influence of different object detection architectures is examined. The results of our empirical study highlight the major effects of forgetting for class-incremental learning. Moreover, we show that domain-incremental learning suffers less from forgetting but is highly dependent on the design of the experiments and choice of architecture.

Paper Citation

in Harvard Style

Witte C., Schuster R., Bukhari S., Trampert P., Stricker D. and Schneider G. (2023). Severity of Catastrophic Forgetting in Object Detection for Autonomous Driving. In Proceedings of the 12th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM, ISBN 978-989-758-626-2, pages 262-269. DOI: 10.5220/0011634500003411

in Bibtex Style

@conference{icpram23,
author={Christian Witte and René Schuster and Syed Bukhari and Patrick Trampert and Didier Stricker and Georg Schneider},
title={Severity of Catastrophic Forgetting in Object Detection for Autonomous Driving},
booktitle={Proceedings of the 12th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,},
year={2023},
pages={262-269},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011634500003411},
isbn={978-989-758-626-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 12th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,
TI - Severity of Catastrophic Forgetting in Object Detection for Autonomous Driving
SN - 978-989-758-626-2
AU - Witte C.
AU - Schuster R.
AU - Bukhari S.
AU - Trampert P.
AU - Stricker D.
AU - Schneider G.
PY - 2023
SP - 262
EP - 269
DO - 10.5220/0011634500003411