FA3D: Fast and Accurate 3D Object Detection

Hamed Habibi Aghdam, Selameab Demilew*, Robert Laganiére, Emil Petriu

*Korrespondierende/r Autor/in für diese Arbeit

Publikation: Beitrag in Buch/Bericht/KonferenzbandBeitrag in einem Konferenzband

Abstract

Fast and accurate detection of objects, in 3D, is one of the critical components in an advanced driver assistance system. In this paper, we aim to develop an accurate 3D object detector that runs in near real-time on low-end embedded systems. We propose an efficient framework that converts raw point cloud into a 3D occupancy cuboid and detects cars using a deep convolutional neural network. Even though the complexity of our proposed model is high, it runs at 7.27 FPS on a Jetson Xavier and at 57.83 FPS on a high-end workstation that is 18 % and 43 % faster than the fastest published method while having a comparable performance with state-of-the-art models on the KITTI dataset. We conduct a comprehensive error analysis on our model and show that two quantities are the principal sources of error among nine predicted attributes. Our source code is available at https://github.com/Selameab/FA3D. © 2020, Springer Nature Switzerland AG.
Originalspracheenglisch
TitelAdvances in Visual Computing - 15th International Symposium, ISVC 2020, Proceedings
Redakteure/-innenGeorge Bebis, Zhaozheng Yin, Edward Kim, Jan Bender, Kartic Subr, Bum Chul Kwon, Jian Zhao, Denis Kalkofen, George Baciu
Seiten397-409
Seitenumfang13
DOIs
PublikationsstatusVeröffentlicht - 7 Dez 2020
Extern publiziertJa
Veranstaltung15th International Symposium on Visual Computing: ISVC 2020 - Virtuell, USA / Vereinigte Staaten
Dauer: 5 Okt 20207 Okt 2020

Publikationsreihe

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Band12509 LNCS
ISSN (Print)0302-9743
ISSN (elektronisch)1611-3349

Konferenz

Konferenz15th International Symposium on Visual Computing
LandUSA / Vereinigte Staaten
OrtVirtuell
Zeitraum5/10/207/10/20

ASJC Scopus subject areas

  • !!Theoretical Computer Science
  • !!Computer Science(all)

Fingerprint

Untersuchen Sie die Forschungsthemen von „FA3D: Fast and Accurate 3D Object Detection“. Zusammen bilden sie einen einzigartigen Fingerprint.

Dieses zitieren