Resumé

Failure in pedestrian detection systems can be extremely crucial, specifically in driverless driving. In this paper, failures in pedestrian detectors are refined by re-evaluating the results of state of the art pedestrian detection systems, via a fully convolutional neural network. The network is trained on a number of datasets which include a custom designed occluded pedestrian dataset to address the problem of occlusion. Results show that when applying the proposed network, detectors can not only maintain their state of the art performance, but they even decrease average false positives rate per image, especially in the case where pedestrians are occluded.
OriginalsprogEngelsk
TitelThe 11th International Conference on Machine Vision
Antal sider8
Vol/bind1104101
ForlagSPIE - International Society for Optical Engineering
Publikationsdato15 mar. 2019
Sider110410I
DOI
StatusUdgivet - 15 mar. 2019
BegivenhedThe 11th International Conference on Machine Vision - Munich, Tyskland
Varighed: 1 nov. 20183 nov. 2018

Konference

KonferenceThe 11th International Conference on Machine Vision
LandTyskland
ByMunich
Periode01/11/201803/11/2018

Fingerprint

Detectors
Neural networks

Citer dette

Apostolopoulos, C., Nasrollahi, K., Yang, M-H. Y., Jahromi, M. N. S., & Moeslund, T. B. (2019). OCCLUSION-AWARE PEDESTRIAN DETECTION. I The 11th International Conference on Machine Vision (Bind 1104101, s. 110410I). SPIE - International Society for Optical Engineering. https://doi.org/10.1117/12.2523107
Apostolopoulos, Christos ; Nasrollahi, Kamal ; Yang, Ming-Hsuan Yang ; Jahromi, Mohammad Naser Sabet ; Moeslund, Thomas B. / OCCLUSION-AWARE PEDESTRIAN DETECTION. The 11th International Conference on Machine Vision. Bind 1104101 SPIE - International Society for Optical Engineering, 2019. s. 110410I
@inproceedings{922252719ae84b5f958bef4932c73ac3,
title = "OCCLUSION-AWARE PEDESTRIAN DETECTION",
abstract = "Failure in pedestrian detection systems can be extremely crucial, specifically in driverless driving. In this paper, failures in pedestrian detectors are refined by re-evaluating the results of state of the art pedestrian detection systems, via a fully convolutional neural network. The network is trained on a number of datasets which include a custom designed occluded pedestrian dataset to address the problem of occlusion. Results show that when applying the proposed network, detectors can not only maintain their state of the art performance, but they even decrease average false positives rate per image, especially in the case where pedestrians are occluded.",
author = "Christos Apostolopoulos and Kamal Nasrollahi and Yang, {Ming-Hsuan Yang} and Jahromi, {Mohammad Naser Sabet} and Moeslund, {Thomas B.}",
year = "2019",
month = "3",
day = "15",
doi = "https://doi.org/10.1117/12.2523107",
language = "English",
volume = "1104101",
pages = "110410I",
booktitle = "The 11th International Conference on Machine Vision",
publisher = "SPIE - International Society for Optical Engineering",
address = "United States",

}

Apostolopoulos, C, Nasrollahi, K, Yang, M-HY, Jahromi, MNS & Moeslund, TB 2019, OCCLUSION-AWARE PEDESTRIAN DETECTION. i The 11th International Conference on Machine Vision. bind 1104101, SPIE - International Society for Optical Engineering, s. 110410I, Munich, Tyskland, 01/11/2018. https://doi.org/10.1117/12.2523107

OCCLUSION-AWARE PEDESTRIAN DETECTION. / Apostolopoulos, Christos; Nasrollahi, Kamal; Yang, Ming-Hsuan Yang; Jahromi, Mohammad Naser Sabet; Moeslund, Thomas B.

The 11th International Conference on Machine Vision. Bind 1104101 SPIE - International Society for Optical Engineering, 2019. s. 110410I.

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

TY - GEN

T1 - OCCLUSION-AWARE PEDESTRIAN DETECTION

AU - Apostolopoulos, Christos

AU - Nasrollahi, Kamal

AU - Yang, Ming-Hsuan Yang

AU - Jahromi, Mohammad Naser Sabet

AU - Moeslund, Thomas B.

PY - 2019/3/15

Y1 - 2019/3/15

N2 - Failure in pedestrian detection systems can be extremely crucial, specifically in driverless driving. In this paper, failures in pedestrian detectors are refined by re-evaluating the results of state of the art pedestrian detection systems, via a fully convolutional neural network. The network is trained on a number of datasets which include a custom designed occluded pedestrian dataset to address the problem of occlusion. Results show that when applying the proposed network, detectors can not only maintain their state of the art performance, but they even decrease average false positives rate per image, especially in the case where pedestrians are occluded.

AB - Failure in pedestrian detection systems can be extremely crucial, specifically in driverless driving. In this paper, failures in pedestrian detectors are refined by re-evaluating the results of state of the art pedestrian detection systems, via a fully convolutional neural network. The network is trained on a number of datasets which include a custom designed occluded pedestrian dataset to address the problem of occlusion. Results show that when applying the proposed network, detectors can not only maintain their state of the art performance, but they even decrease average false positives rate per image, especially in the case where pedestrians are occluded.

U2 - https://doi.org/10.1117/12.2523107

DO - https://doi.org/10.1117/12.2523107

M3 - Article in proceeding

VL - 1104101

SP - 110410I

BT - The 11th International Conference on Machine Vision

PB - SPIE - International Society for Optical Engineering

ER -

Apostolopoulos C, Nasrollahi K, Yang M-HY, Jahromi MNS, Moeslund TB. OCCLUSION-AWARE PEDESTRIAN DETECTION. I The 11th International Conference on Machine Vision. Bind 1104101. SPIE - International Society for Optical Engineering. 2019. s. 110410I https://doi.org/10.1117/12.2523107