Classifying objects in LWIR imagery via CNNs

Iain Rodger; Barry Connor; Neil M. Robertson

doi:10.1117/12.2241858

Classifying objects in LWIR imagery via CNNs

Iain Rodger, Barry Connor, Neil M. Robertson

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

20 Citations (Scopus)

2284 Downloads (Pure)

Abstract

The aim of the presented work is to demonstrate enhanced target recognition and improved false alarm rates for a mid to long range detection system, utilising a Long Wave Infrared (LWIR) sensor. By exploiting high quality thermal image data and recent techniques in machine learning, the system can provide automatic target recognition capabilities. A Convolutional Neural Network (CNN) is trained and the classifier achieves an overall accuracy of > 95% for 6 object classes related to land defence. While the highly accurate CNN struggles to recognise long range target classes, due to low signal quality, robust target discrimination is achieved for challenging candidates. The overall performance of the methodology presented is assessed using human ground truth information, generating classifier evaluation metrics for thermal image sequences.

Original language	English
Title of host publication	Proc. SPIE: Electro-Optical and Infrared Systems: Technology and Applications XIII
Pages	99870-99884
Number of pages	14
Volume	9987
DOIs	https://doi.org/10.1117/12.2241858
Publication status	Published - 21 Oct 2016

Bibliographical note

Winner of Best Student Paper prize

Access to Document

10.1117/12.2241858Licence: Unspecified

Classifying Objects in LWIR Imagery via CNNs
Copyright 2016 Society of Photo-Optical Instrumentation Engineers (SPIE). One print or electronic copy may be made for personal use only. Systematic reproduction and distribution, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited.
Accepted author manuscript, 3.45 MBLicence: Unspecified

Cite this

@inproceedings{4710779f415641beb364d67ac28cae0f,

title = "Classifying objects in LWIR imagery via CNNs",

abstract = "The aim of the presented work is to demonstrate enhanced target recognition and improved false alarm rates for a mid to long range detection system, utilising a Long Wave Infrared (LWIR) sensor. By exploiting high quality thermal image data and recent techniques in machine learning, the system can provide automatic target recognition capabilities. A Convolutional Neural Network (CNN) is trained and the classifier achieves an overall accuracy of > 95% for 6 object classes related to land defence. While the highly accurate CNN struggles to recognise long range target classes, due to low signal quality, robust target discrimination is achieved for challenging candidates. The overall performance of the methodology presented is assessed using human ground truth information, generating classifier evaluation metrics for thermal image sequences.",

author = "Iain Rodger and Barry Connor and Robertson, {Neil M.}",

note = "Winner of Best Student Paper prize",

year = "2016",

month = oct,

day = "21",

doi = "10.1117/12.2241858",

language = "English",

volume = "9987",

pages = "99870--99884",

booktitle = "Proc. SPIE: Electro-Optical and Infrared Systems: Technology and Applications XIII",

}

TY - GEN

T1 - Classifying objects in LWIR imagery via CNNs

AU - Rodger, Iain

AU - Connor, Barry

AU - Robertson, Neil M.

N1 - Winner of Best Student Paper prize

PY - 2016/10/21

Y1 - 2016/10/21

N2 - The aim of the presented work is to demonstrate enhanced target recognition and improved false alarm rates for a mid to long range detection system, utilising a Long Wave Infrared (LWIR) sensor. By exploiting high quality thermal image data and recent techniques in machine learning, the system can provide automatic target recognition capabilities. A Convolutional Neural Network (CNN) is trained and the classifier achieves an overall accuracy of > 95% for 6 object classes related to land defence. While the highly accurate CNN struggles to recognise long range target classes, due to low signal quality, robust target discrimination is achieved for challenging candidates. The overall performance of the methodology presented is assessed using human ground truth information, generating classifier evaluation metrics for thermal image sequences.

AB - The aim of the presented work is to demonstrate enhanced target recognition and improved false alarm rates for a mid to long range detection system, utilising a Long Wave Infrared (LWIR) sensor. By exploiting high quality thermal image data and recent techniques in machine learning, the system can provide automatic target recognition capabilities. A Convolutional Neural Network (CNN) is trained and the classifier achieves an overall accuracy of > 95% for 6 object classes related to land defence. While the highly accurate CNN struggles to recognise long range target classes, due to low signal quality, robust target discrimination is achieved for challenging candidates. The overall performance of the methodology presented is assessed using human ground truth information, generating classifier evaluation metrics for thermal image sequences.

U2 - 10.1117/12.2241858

DO - 10.1117/12.2241858

M3 - Conference contribution

VL - 9987

SP - 99870

EP - 99884

BT - Proc. SPIE: Electro-Optical and Infrared Systems: Technology and Applications XIII

ER -

Classifying objects in LWIR imagery via CNNs

Abstract

Bibliographical note

Access to Document

Fingerprint

Cite this