Research papers · Posted 2020.11.03 · Publication / acceptance: author-reported
Out-of-Distribution Detection for Automotive Perception
Out-of-Distribution Detection for Automotive Perception
Overview
The paper proposes automotive out-of-distribution detection without requiring OOD training examples or additional inference cost. Real-world automotive datasets evaluate the method against confidence-estimation limitations associated with softmax-based classification.
Publication and version information
Publication, acceptance or venue information is author-reported on arXiv. The exact source wording is preserved below.
Author-supplied source information
Author-supplied information on arXiv.
6 pages, 4 figures, paper accepted at Intelligent Transportation Systems Conference (ITSC) 2021
Version: arXiv v2 — manuscript revision