The embeddability of lane detection algorithms on heterogeneous architectures - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2015

The embeddability of lane detection algorithms on heterogeneous architectures

Résumé

Lane detection plays a crucial role for Advanced Driver As-sitance System (ADAS) or autonomous driving applications. Literature shows a lot of lane detection algorithms can work in real time with good results. However, they require much computer processing and cannot be embedded in a vehicle ECU without deep software optimizations. In this paper, we discuss the embeddability of lane detection algorithms by comparing state-of-the-art algorithms in terms of functional performance and computational timing. We identify what essential parts of lane detection are time consuming, and show these parts can be computed in real time on embedded systems.
Fichier principal
Vignette du fichier
Saussard_ICIP2015.pdf (144.19 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01244409 , version 1 (15-12-2015)

Identifiants

Citer

Romain Saussard, Boubker Bouzid, Marius Vasiliu, Roger Reynaud. The embeddability of lane detection algorithms on heterogeneous architectures. IEEE International Conference on Image Processing (ICIP), Sep 2015, Quebec City, Canada. ⟨10.1109/ICIP.2015.7351697⟩. ⟨hal-01244409⟩
99 Consultations
536 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More