Real-Time Multi-Scale Pedestrian Detection for Driver Assistance Systems

Maryam Hemmati (1) , Morteza Biglari-Abhari (1) , Smail Niar (2) , Stevan Berber (1)

1 ECE - Department of Electrical and Computer Engineering [Auckland ]
2 LAMIH - Laboratoire d'Automatique, de Mécanique et d'Informatique industrielles et Humaines - UMR 8201

Pedestrian detection is one of the most challenging and vital tasks of driver assistance systems (DAS). Among several algorithms developed for human detection, histogram of oriented gradients (HOG) followed by support vector machine (SVM) has shown the most promising results. This paper presents a hardware accelerator for real-time pedestrian detection at different scales to fulfill the real-time requirements of DAS. It proposes an algorithmic modification to the conventional multi-scale object detection by means of HOG+SVM to increase the throughput and maintain the accuracy reasonably high. Our hardware accelerator detects pedestrians at the rate of 60 fps for HDTV (1080×1920) frame.

Mots clés: Feature extraction; Support vector machines; Real-time systems; Object detection; Hardware; Vehicles

Domaines: Informatique [cs]

Maryam Hemmati, Morteza Biglari-Abhari, Smail Niar, Stevan Berber. Real-Time Multi-Scale Pedestrian Detection for Driver Assistance Systems. Design Automation Conference (IEEE-ACM DAC), Jun 2017, Austin, TX, United States. pp.1-6, ⟨10.1145/3061639.3062308⟩. ⟨hal-03388354⟩