Improving a real-time object detector with compact temporal information

Martin Ahrnbom; Morten Bornø Jensen; Karl Åström; Mikael Nilsson; Håkan Ardö; Thomas Moeslund

doi:10.1109/ICCVW.2017.31

Improving a real-time object detector with compact temporal information

Martin Ahrnbom, Morten Bornø Jensen, Karl Åström, Mikael Nilsson, Håkan Ardö, Thomas Moeslund

Aalborg University

Research output: Chapter in Book/Report/Conference proceeding › Paper in conference proceeding › peer-review

Abstract

Neural networks designed for real-time object detection
have recently improved significantly, but in practice, look-
ing at only a single RGB image at the time may not be ideal.
For example, when detecting objects in videos, a foreground
detection algorithm can be used to obtain compact temporal
data, which can be fed into a neural network alongside RGB
images. We propose an approach for doing this, based on
an existing object detector, that re-uses pretrained weights
for the processing of RGB images. The neural network was
tested on the VIRAT dataset with annotations for object de-
tection, a problem this approach is well suited for. The ac-
curacy was found to improve significantly (up to 66%), with
a roughly 40% increase in computational time.

Original language	English
Title of host publication	International Conference on Computer Vision Workshops, 2017
Subtitle of host publication	Computer Vision for Road Scene Understanding and Autonomous Driving Workshop
Publisher	IEEE - Institute of Electrical and Electronics Engineers Inc.
Pages	190-197
Number of pages	8
DOIs	https://doi.org/10.1109/ICCVW.2017.31
Publication status	Published - 2018 Jan 19
Event	16th IEEE International Conference on Computer Vision Workshops, ICCVW 2017 - Venice, Italy Duration: 2017 Oct 22 → 2017 Oct 29

Conference

Conference	16th IEEE International Conference on Computer Vision Workshops, ICCVW 2017
Country/Territory	Italy
City	Venice
Period	2017/10/22 → 2017/10/29

Subject classification (UKÄ)

Computer Vision and Robotics (Autonomous Systems)

Access to Document

10.1109/ICCVW.2017.31

http://openaccess.thecvf.com/content_ICCV_2017_workshops/papers/w3/Ahrnbom_Improving_a_Real-Time_ICCV_2017_paper.pdf

1 Active
2 Finished

AI Lund: Lund University AI Research
Åström, K. (Project coordinator), Malec, J. (Project coordinator), Larsson, S. (Project coordinator), Ohlsson, M. (Project coordinator), Balkenius, C. (Project coordinator), Dutceac Segesten, A. (Project coordinator), Haider, J. (Project coordinator), Willim, R. (Project coordinator), Ledendal, J. (Project coordinator), Aits, S. (Project coordinator), Hedlund, M. (Researcher), Wisbrant, J. (Project communication officer), Heiberg, E. (PI), Topp, E. A. (Researcher), Janneck, J. (Researcher), Klang, M. (Researcher), Brinck, I. (Researcher), Sundin, O. (Researcher), Westin, E. (Administrator), Blom Lussi, E. (Project coordinator) & Axhamn, J. (Project coordinator)
2018/01/01 → …
Project: Network
Semantic Mapping and Visual Navigation for Smart Robots
Åström, K. (PI), Sminchisescu, C. (PI), Kahl, F. (PI), Robertsson, A. (PI), Flood, G. (Research student), Priisalu, M. (Research student), Greiff, M. (Research student) & Sun, Z. (Researcher)
2016/07/01 → 2022/06/30
Project: Research
InDeV: In-Depth understanding of accident causation for Vulnerable road users
Laureshyn, A. (Project coordinator), Varhelyi, A. (Researcher), Johnsson, C. (Research student), D'agostino, C. (Researcher), Svensson, Å. (Researcher) & Norén, H. (Research assistant)
2014/05/01 → 2018/10/31
Project: Research

Cite this

Ahrnbom, M., Bornø Jensen, M., Åström, K., Nilsson, M., Ardö, H., & Moeslund, T. (2018). Improving a real-time object detector with compact temporal information. In International Conference on Computer Vision Workshops, 2017: Computer Vision for Road Scene Understanding and Autonomous Driving Workshop (pp. 190-197). IEEE - Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICCVW.2017.31

@inproceedings{41030db4c01e4609be4863157ea594d4,

title = "Improving a real-time object detector with compact temporal information",

abstract = "Neural networks designed for real-time object detectionhave recently improved significantly, but in practice, look-ing at only a single RGB image at the time may not be ideal.For example, when detecting objects in videos, a foregrounddetection algorithm can be used to obtain compact temporaldata, which can be fed into a neural network alongside RGBimages. We propose an approach for doing this, based onan existing object detector, that re-uses pretrained weightsfor the processing of RGB images. The neural network wastested on the VIRAT dataset with annotations for object de-tection, a problem this approach is well suited for. The ac-curacy was found to improve significantly (up to 66%), witha roughly 40% increase in computational time.",

author = "Martin Ahrnbom and {Born{\o} Jensen}, Morten and Karl {\AA}str{\"o}m and Mikael Nilsson and H{\aa}kan Ard{\"o} and Thomas Moeslund",

year = "2018",

month = jan,

day = "19",

doi = "10.1109/ICCVW.2017.31",

language = "English",

pages = "190--197",

booktitle = "International Conference on Computer Vision Workshops, 2017",

publisher = "IEEE - Institute of Electrical and Electronics Engineers Inc.",

address = "United States",

note = "16th IEEE International Conference on Computer Vision Workshops, ICCVW 2017 ; Conference date: 22-10-2017 Through 29-10-2017",

}

Ahrnbom, M, Bornø Jensen, M, Åström, K , Nilsson, M , Ardö, H & Moeslund, T 2018, Improving a real-time object detector with compact temporal information. in International Conference on Computer Vision Workshops, 2017: Computer Vision for Road Scene Understanding and Autonomous Driving Workshop. IEEE - Institute of Electrical and Electronics Engineers Inc., pp. 190-197, 16th IEEE International Conference on Computer Vision Workshops, ICCVW 2017, Venice, Italy, 2017/10/22. https://doi.org/10.1109/ICCVW.2017.31

Improving a real-time object detector with compact temporal information. / Ahrnbom, Martin; Bornø Jensen, Morten; Åström, Karl et al.
International Conference on Computer Vision Workshops, 2017: Computer Vision for Road Scene Understanding and Autonomous Driving Workshop. IEEE - Institute of Electrical and Electronics Engineers Inc., 2018. p. 190-197.

Research output: Chapter in Book/Report/Conference proceeding › Paper in conference proceeding › peer-review

TY - GEN

T1 - Improving a real-time object detector with compact temporal information

AU - Ahrnbom, Martin

AU - Bornø Jensen, Morten

AU - Åström, Karl

AU - Nilsson, Mikael

AU - Ardö, Håkan

AU - Moeslund, Thomas

PY - 2018/1/19

Y1 - 2018/1/19

N2 - Neural networks designed for real-time object detectionhave recently improved significantly, but in practice, look-ing at only a single RGB image at the time may not be ideal.For example, when detecting objects in videos, a foregrounddetection algorithm can be used to obtain compact temporaldata, which can be fed into a neural network alongside RGBimages. We propose an approach for doing this, based onan existing object detector, that re-uses pretrained weightsfor the processing of RGB images. The neural network wastested on the VIRAT dataset with annotations for object de-tection, a problem this approach is well suited for. The ac-curacy was found to improve significantly (up to 66%), witha roughly 40% increase in computational time.

AB - Neural networks designed for real-time object detectionhave recently improved significantly, but in practice, look-ing at only a single RGB image at the time may not be ideal.For example, when detecting objects in videos, a foregrounddetection algorithm can be used to obtain compact temporaldata, which can be fed into a neural network alongside RGBimages. We propose an approach for doing this, based onan existing object detector, that re-uses pretrained weightsfor the processing of RGB images. The neural network wastested on the VIRAT dataset with annotations for object de-tection, a problem this approach is well suited for. The ac-curacy was found to improve significantly (up to 66%), witha roughly 40% increase in computational time.

U2 - 10.1109/ICCVW.2017.31

DO - 10.1109/ICCVW.2017.31

M3 - Paper in conference proceeding

SP - 190

EP - 197

BT - International Conference on Computer Vision Workshops, 2017

PB - IEEE - Institute of Electrical and Electronics Engineers Inc.

T2 - 16th IEEE International Conference on Computer Vision Workshops, ICCVW 2017

Y2 - 22 October 2017 through 29 October 2017

ER -

Ahrnbom M, Bornø Jensen M, Åström K , Nilsson M , Ardö H, Moeslund T. Improving a real-time object detector with compact temporal information. In International Conference on Computer Vision Workshops, 2017: Computer Vision for Road Scene Understanding and Autonomous Driving Workshop. IEEE - Institute of Electrical and Electronics Engineers Inc. 2018. p. 190-197 doi: 10.1109/ICCVW.2017.31

Improving a real-time object detector with compact temporal information

Abstract

Conference

Subject classification (UKÄ)

Access to Document

Fingerprint

Projects

AI Lund: Lund University AI Research

Semantic Mapping and Visual Navigation for Smart Robots

InDeV: In-Depth understanding of accident causation for Vulnerable road users

Cite this