Reconstruction-free inference on compressive measurements

Suhas Lohit, Kuldeep Kulkarni, Pavan Turaga, Jian Wang, Aswin C. Sankaranarayanan

Research output: Chapter in Book/Report/Conference proceedingConference contribution

10 Citations (Scopus)

Abstract

Spatial-multiplexing cameras have emerged as a promising alternative to classical imaging devices, often enabling acquisition of 'more for less'. One popular architecture for spatial multiplexing is the single-pixel camera (SPC), which acquires coded measurements of the scene with pseudo-random spatial masks. Significant theoretical developments over the past few years provide a means for reconstruction of the original imagery from coded measurements at sub-Nyquist sampling rates. Yet, accurate reconstruction generally requires high measurement rates and high signal-to-noise ratios. In this paper, we enquire if one can perform high-level visual inference problems (e.g. face recognition or action recognition) from compressive cameras without the need for image reconstruction. This is an interesting question since in many practical scenarios, our goals extend beyond image reconstruction. However, most inference tasks often require non-linear features and it is not clear how to extract such features directly from compressed measurements. In this paper, we show that one can extract nontrivial correlational features directly without reconstruction of the imagery. As a specific example, we consider the problem of face recognition beyond the visible spectrum e.g in the short-wave infra-red region (SWIR)-where pixels are expensive. We base our framework on smashed filters which suggests that inner-products between high-dimensional signals can be computed in the compressive domain to a high degree of accuracy. We collect a new face image dataset of 30 subjects, obtained using an SPC. Using face recognition as an example, we show that one can indeed perform reconstruction-free inference with a very small loss of accuracy at very high compression ratios of 100 and more.

Original languageEnglish (US)
Title of host publicationIEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops
PublisherIEEE Computer Society
Pages16-24
Number of pages9
Volume2015-October
ISBN (Print)9781467367592
DOIs
StatePublished - Oct 19 2015
EventIEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2015 - Boston, United States
Duration: Jun 7 2015Jun 12 2015

Other

OtherIEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2015
CountryUnited States
CityBoston
Period6/7/156/12/15

Fingerprint

Face recognition
Cameras
Pixels
Image reconstruction
Multiplexing
Masks
Signal to noise ratio
Sampling
Infrared radiation
Imaging techniques

Keywords

  • Correlation
  • Databases
  • Face
  • Face recognition
  • Feature extraction
  • Image coding
  • Image reconstruction

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering

Cite this

Lohit, S., Kulkarni, K., Turaga, P., Wang, J., & Sankaranarayanan, A. C. (2015). Reconstruction-free inference on compressive measurements. In IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (Vol. 2015-October, pp. 16-24). [7301371] IEEE Computer Society. https://doi.org/10.1109/CVPRW.2015.7301371

Reconstruction-free inference on compressive measurements. / Lohit, Suhas; Kulkarni, Kuldeep; Turaga, Pavan; Wang, Jian; Sankaranarayanan, Aswin C.

IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops. Vol. 2015-October IEEE Computer Society, 2015. p. 16-24 7301371.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Lohit, S, Kulkarni, K, Turaga, P, Wang, J & Sankaranarayanan, AC 2015, Reconstruction-free inference on compressive measurements. in IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops. vol. 2015-October, 7301371, IEEE Computer Society, pp. 16-24, IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2015, Boston, United States, 6/7/15. https://doi.org/10.1109/CVPRW.2015.7301371
Lohit S, Kulkarni K, Turaga P, Wang J, Sankaranarayanan AC. Reconstruction-free inference on compressive measurements. In IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops. Vol. 2015-October. IEEE Computer Society. 2015. p. 16-24. 7301371 https://doi.org/10.1109/CVPRW.2015.7301371
Lohit, Suhas ; Kulkarni, Kuldeep ; Turaga, Pavan ; Wang, Jian ; Sankaranarayanan, Aswin C. / Reconstruction-free inference on compressive measurements. IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops. Vol. 2015-October IEEE Computer Society, 2015. pp. 16-24
@inproceedings{fde5e409d52b4b6cbc6b920897bad8c5,
title = "Reconstruction-free inference on compressive measurements",
abstract = "Spatial-multiplexing cameras have emerged as a promising alternative to classical imaging devices, often enabling acquisition of 'more for less'. One popular architecture for spatial multiplexing is the single-pixel camera (SPC), which acquires coded measurements of the scene with pseudo-random spatial masks. Significant theoretical developments over the past few years provide a means for reconstruction of the original imagery from coded measurements at sub-Nyquist sampling rates. Yet, accurate reconstruction generally requires high measurement rates and high signal-to-noise ratios. In this paper, we enquire if one can perform high-level visual inference problems (e.g. face recognition or action recognition) from compressive cameras without the need for image reconstruction. This is an interesting question since in many practical scenarios, our goals extend beyond image reconstruction. However, most inference tasks often require non-linear features and it is not clear how to extract such features directly from compressed measurements. In this paper, we show that one can extract nontrivial correlational features directly without reconstruction of the imagery. As a specific example, we consider the problem of face recognition beyond the visible spectrum e.g in the short-wave infra-red region (SWIR)-where pixels are expensive. We base our framework on smashed filters which suggests that inner-products between high-dimensional signals can be computed in the compressive domain to a high degree of accuracy. We collect a new face image dataset of 30 subjects, obtained using an SPC. Using face recognition as an example, we show that one can indeed perform reconstruction-free inference with a very small loss of accuracy at very high compression ratios of 100 and more.",
keywords = "Correlation, Databases, Face, Face recognition, Feature extraction, Image coding, Image reconstruction",
author = "Suhas Lohit and Kuldeep Kulkarni and Pavan Turaga and Jian Wang and Sankaranarayanan, {Aswin C.}",
year = "2015",
month = "10",
day = "19",
doi = "10.1109/CVPRW.2015.7301371",
language = "English (US)",
isbn = "9781467367592",
volume = "2015-October",
pages = "16--24",
booktitle = "IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops",
publisher = "IEEE Computer Society",

}

TY - GEN

T1 - Reconstruction-free inference on compressive measurements

AU - Lohit, Suhas

AU - Kulkarni, Kuldeep

AU - Turaga, Pavan

AU - Wang, Jian

AU - Sankaranarayanan, Aswin C.

PY - 2015/10/19

Y1 - 2015/10/19

N2 - Spatial-multiplexing cameras have emerged as a promising alternative to classical imaging devices, often enabling acquisition of 'more for less'. One popular architecture for spatial multiplexing is the single-pixel camera (SPC), which acquires coded measurements of the scene with pseudo-random spatial masks. Significant theoretical developments over the past few years provide a means for reconstruction of the original imagery from coded measurements at sub-Nyquist sampling rates. Yet, accurate reconstruction generally requires high measurement rates and high signal-to-noise ratios. In this paper, we enquire if one can perform high-level visual inference problems (e.g. face recognition or action recognition) from compressive cameras without the need for image reconstruction. This is an interesting question since in many practical scenarios, our goals extend beyond image reconstruction. However, most inference tasks often require non-linear features and it is not clear how to extract such features directly from compressed measurements. In this paper, we show that one can extract nontrivial correlational features directly without reconstruction of the imagery. As a specific example, we consider the problem of face recognition beyond the visible spectrum e.g in the short-wave infra-red region (SWIR)-where pixels are expensive. We base our framework on smashed filters which suggests that inner-products between high-dimensional signals can be computed in the compressive domain to a high degree of accuracy. We collect a new face image dataset of 30 subjects, obtained using an SPC. Using face recognition as an example, we show that one can indeed perform reconstruction-free inference with a very small loss of accuracy at very high compression ratios of 100 and more.

AB - Spatial-multiplexing cameras have emerged as a promising alternative to classical imaging devices, often enabling acquisition of 'more for less'. One popular architecture for spatial multiplexing is the single-pixel camera (SPC), which acquires coded measurements of the scene with pseudo-random spatial masks. Significant theoretical developments over the past few years provide a means for reconstruction of the original imagery from coded measurements at sub-Nyquist sampling rates. Yet, accurate reconstruction generally requires high measurement rates and high signal-to-noise ratios. In this paper, we enquire if one can perform high-level visual inference problems (e.g. face recognition or action recognition) from compressive cameras without the need for image reconstruction. This is an interesting question since in many practical scenarios, our goals extend beyond image reconstruction. However, most inference tasks often require non-linear features and it is not clear how to extract such features directly from compressed measurements. In this paper, we show that one can extract nontrivial correlational features directly without reconstruction of the imagery. As a specific example, we consider the problem of face recognition beyond the visible spectrum e.g in the short-wave infra-red region (SWIR)-where pixels are expensive. We base our framework on smashed filters which suggests that inner-products between high-dimensional signals can be computed in the compressive domain to a high degree of accuracy. We collect a new face image dataset of 30 subjects, obtained using an SPC. Using face recognition as an example, we show that one can indeed perform reconstruction-free inference with a very small loss of accuracy at very high compression ratios of 100 and more.

KW - Correlation

KW - Databases

KW - Face

KW - Face recognition

KW - Feature extraction

KW - Image coding

KW - Image reconstruction

UR - http://www.scopus.com/inward/record.url?scp=84952004606&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84952004606&partnerID=8YFLogxK

U2 - 10.1109/CVPRW.2015.7301371

DO - 10.1109/CVPRW.2015.7301371

M3 - Conference contribution

AN - SCOPUS:84952004606

SN - 9781467367592

VL - 2015-October

SP - 16

EP - 24

BT - IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops

PB - IEEE Computer Society

ER -