Convolutional Networks for Shape from Light Field

Stefan Heber, Thomas Pock

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

Convolutional Neural Networks (CNNs) have recently been successfully applied to various Computer Vision (CV) applications. In this paper we utilize CNNs to predict depth information for given Light Field (LF) data. The proposed method learns an end-to-end mapping between the 4D light field and a representation of the corresponding 4D depth field in terms of 2D hyperplane orientations. The obtained prediction is then further refined in a post processing step by applying a higher-order regularization.
Existing LF datasets are not sufficient for the purpose of the training scheme tackled in this paper. This is mainly due to the fact that the ground truth depth of existing datasets is inaccurate and/or the datasets are limited to a small number of LFs. This made it necessary to generate a new synthetic LF dataset, which is based on the raytracing software POV-Ray. This new dataset provides floating point accurate ground truth depth fields, and due to a random scene generator the dataset can be scaled as required.
Original languageEnglish
Title of host publicationIEEE Conference on Computer Vision and Pattern Recognition (2016)
Publication statusPublished - 2016

Fingerprint

Neural networks
Computer vision
Processing

Cite this

Heber, S., & Pock, T. (2016). Convolutional Networks for Shape from Light Field. In IEEE Conference on Computer Vision and Pattern Recognition (2016)

Convolutional Networks for Shape from Light Field. / Heber, Stefan; Pock, Thomas.

IEEE Conference on Computer Vision and Pattern Recognition (2016). 2016.

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Heber, S & Pock, T 2016, Convolutional Networks for Shape from Light Field. in IEEE Conference on Computer Vision and Pattern Recognition (2016).
Heber S, Pock T. Convolutional Networks for Shape from Light Field. In IEEE Conference on Computer Vision and Pattern Recognition (2016). 2016
Heber, Stefan ; Pock, Thomas. / Convolutional Networks for Shape from Light Field. IEEE Conference on Computer Vision and Pattern Recognition (2016). 2016.
@inproceedings{297214f934db423f94c06e7840abf323,
title = "Convolutional Networks for Shape from Light Field",
abstract = "Convolutional Neural Networks (CNNs) have recently been successfully applied to various Computer Vision (CV) applications. In this paper we utilize CNNs to predict depth information for given Light Field (LF) data. The proposed method learns an end-to-end mapping between the 4D light field and a representation of the corresponding 4D depth field in terms of 2D hyperplane orientations. The obtained prediction is then further refined in a post processing step by applying a higher-order regularization.Existing LF datasets are not sufficient for the purpose of the training scheme tackled in this paper. This is mainly due to the fact that the ground truth depth of existing datasets is inaccurate and/or the datasets are limited to a small number of LFs. This made it necessary to generate a new synthetic LF dataset, which is based on the raytracing software POV-Ray. This new dataset provides floating point accurate ground truth depth fields, and due to a random scene generator the dataset can be scaled as required.",
author = "Stefan Heber and Thomas Pock",
year = "2016",
language = "English",
booktitle = "IEEE Conference on Computer Vision and Pattern Recognition (2016)",

}

TY - GEN

T1 - Convolutional Networks for Shape from Light Field

AU - Heber, Stefan

AU - Pock, Thomas

PY - 2016

Y1 - 2016

N2 - Convolutional Neural Networks (CNNs) have recently been successfully applied to various Computer Vision (CV) applications. In this paper we utilize CNNs to predict depth information for given Light Field (LF) data. The proposed method learns an end-to-end mapping between the 4D light field and a representation of the corresponding 4D depth field in terms of 2D hyperplane orientations. The obtained prediction is then further refined in a post processing step by applying a higher-order regularization.Existing LF datasets are not sufficient for the purpose of the training scheme tackled in this paper. This is mainly due to the fact that the ground truth depth of existing datasets is inaccurate and/or the datasets are limited to a small number of LFs. This made it necessary to generate a new synthetic LF dataset, which is based on the raytracing software POV-Ray. This new dataset provides floating point accurate ground truth depth fields, and due to a random scene generator the dataset can be scaled as required.

AB - Convolutional Neural Networks (CNNs) have recently been successfully applied to various Computer Vision (CV) applications. In this paper we utilize CNNs to predict depth information for given Light Field (LF) data. The proposed method learns an end-to-end mapping between the 4D light field and a representation of the corresponding 4D depth field in terms of 2D hyperplane orientations. The obtained prediction is then further refined in a post processing step by applying a higher-order regularization.Existing LF datasets are not sufficient for the purpose of the training scheme tackled in this paper. This is mainly due to the fact that the ground truth depth of existing datasets is inaccurate and/or the datasets are limited to a small number of LFs. This made it necessary to generate a new synthetic LF dataset, which is based on the raytracing software POV-Ray. This new dataset provides floating point accurate ground truth depth fields, and due to a random scene generator the dataset can be scaled as required.

M3 - Conference contribution

BT - IEEE Conference on Computer Vision and Pattern Recognition (2016)

ER -