«He who knows most,

doubts most.»

My name is Eduardo Pérez Pellitero. I am a postdoctoral researcher at the Empirical Inference department of the Max Planck Institute for Intelligent Systems, directed by Bernhard Schölkopf, more specifically working within the Computational Imaging Group.

Before this, I spent some time having fun and doing my PhD somewhere in between the TNT Lab of Leibnitz Universität Hannover and Technicolor R&I, where I did research on manifold learning for Super Resolution applications.

My research interests include super-resolution, machine learning and more specifically deep learning. In a broader sense, I am interested in any model that allows us to dissect, analyze and play around with the digital image formation process.

Apart from research, some of the things I enjoy most in my life are: playing the electric bass and practising historical fencing.

2018 First year as a postdoc at the Max Planck Institute

Photorealistic Video Super Resolution

Eduardo Pérez-Pellitero, Mehdi S. M. Sajjadi, Michael Hirsch and Bernhard Schölkopf

Workshop and Challenge on Perceptual Image Restoration and Manipulation (PIRM), ECCV 2018 in Munich 
@conference {Perezpellitero2016_2,
  author = {Eduardo Pérez-Pellitero and Mehdi S. M. Sajjadi and Michael Hirsch and Bernhard Schölkopf},
  title = {Photorealistic Video Super Resolution},
  booktitle = {ECCV Workshop (PIRM)},
  year = {2018},
}

With the advent of perceptual loss functions, new possibilities in super-resolution have emerged, and we currently have models that successfully generate near-photorealistic high-resolution images from their low-resolution observations. Up to now, however, such approaches have been exclusively limited to single image super-resolution. The application of perceptual loss functions on video processing still entails several challenges, mostly related to the lack of temporal consistency of the generated images, i.e., flickering artifacts. In this work, we present a novel adversarial recurrent network for video upscaling that is able to produce realistic textures in a temporally consistent way. The proposed architecture naturally leverages information from previous frames due to its recurrent architecture, i.e. the input to the generator is composed of the low-resolution image and, additionally, the warped output of the network at the previous step. We also propose an additional loss function to further reinforce temporal consistency in the generated sequences. The experimental validation of our algorithm shows the effectiveness of our approach which obtains competitive samples in terms of perceptual quality with improved temporal consistency.


2017 Finished my PhD!

Manifold Learning for Super Resolution

Eduardo Pérez-Pellitero

PhD Dissertation, defended the 21st of February, 2017.
@phdthesis{PerezPellitero2017,
    title    = {Manifold Learning for Super Resolution},
    school   = {Leibniz Universit\"at Hannover},
    author   = {Eduardo P\'erez-Pellitero},
    year     = {2017}, 
}

The development pace of high-resolution displays has been so fast in the recent years that many images acquired with low-end capture devices are already outdated or will be shortly in time. Super Resolution is central to match the resolution of the already existing image content to that of current and future high resolution displays and applications. This dissertation is focused on learning how to upscale images from the statistics of natural images.


2016

PSyCo: Manifold Span Reduction for Super Resolution

Eduardo Pérez-Pellitero, Jordi Salvador, Javier Ruiz-Hidalgo, Bodo Rosenhahn.

IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas.
@inproceedings {Perezpellitero2016_2,
  author = {P\'erez-Pellitero, E. and Salvador, J. and Ruiz-Hidalgo, J. and Rosenhahn, B.},
  title = {PSyCo: Manifold Span Reduction for Super Resolution},
  booktitle = {Proc. {IEEE}  Conference on Computer Vision and Pattern Recognition},
  year = {2016},
}

In this paper we present a novel regression-based SR algorithm that benefits from an extended knowledge of the structure of both manifolds. We propose a transform that collapses the 16 variations induced from the dihedral group of transforms (i.e. rotations, vertical and horizontal reflections) and antipodality (i.e. diametrically opposed points in the unitary sphere) into a single primitive. The key idea of our transform is to study the different dihedral elements as a group of symmetries within the high-dimensional manifold. The experimental validation of our algorithm shows the effectiveness of our approach, which obtains competitive quality with a dictionary of as little as 32 atoms (reducing other methods' dictionaries by at least a factor of 32) and further pushing the state-of-the-art with a 1024 atoms dictionary.

Source Code Available!

Antipodally Invariant Metrics For Fast Regression-based Super-Resolution

Eduardo Pérez-Pellitero, Jordi Salvador, Javier Ruiz-Hidalgo, Bodo Rosenhahn.

IEEE Transactions on Image Processing.
Impact Factor (2015): 3.625
@ARTICLE{PerezPellitero_2016_3,
author={E. Pérez-Pellitero and J. Salvador and J. Ruiz-Hidalgo and B. Rosenhahn},
journal={IEEE Transactions on Image Processing},
title={Antipodally Invariant Metrics for Fast Regression-Based Super-Resolution},
year={2016},
volume={25},
number={6},
pages={2456-2468},
month={June},
}

In this paper we present a very fast regression-based algorithm which builds on densely populated anchored neighborhoods and sublinear search structures. Even though we validate the benefits of using antipodally invariant metrics, most of the binary splits use Euclidean distance, which does not handle antipodes optimally. In order to benefit from both worlds, we propose a simple yet effective Antipodally Invariant Transform (AIT) that can be easily included in the Euclidean distance calculation. We modify the original Spherical Hashing algorithm with this metric in our Antipodally Invariant Spherical Hashing scheme, obtaining the same performance as a pure antipodally invariant metric. We round up our contributions with a novel feature transform that obtains a better coarse approximation of the input image thanks to Iterative Back Projection.

Project website (code available)

Half Hypersphere Confinement for Piecewise Linear Regression

Eduardo Pérez-Pellitero, Jordi Salvador, Javier Ruiz-Hidalgo, Bodo Rosenhahn.

IEEE Winter Conference on Applications of Computer Vision, Lake Placid, NY.
@inproceedings {Perezpellitero2016,
  author = {P\'erez-Pellitero, E. and Salvador, J. and Ruiz-Hidalgo, J. and Rosenhahn, B.},
  title = {Half Hypersphere Confinement for Piecewise Linear Regression},
  booktitle = {Proc. {IEEE}  Winter Conference on Applications of Computer Vision},
  year = {2016},
}

In this paper we study the characteristics of the metrics best suited for the piecewise regression algorithms, in which comparisons are usually made between normalized vectors that lie on the unitary hypersphere. Even though Euclidean distance has been widely used for this purpose, it is suboptimal since it does not handle antipodal points (i.e. diametrically opposite points) properly. Therefore, we propose the usage of antipodally invariant metrics and introduce the Half Hypersphere Confinement (HHC), a fast alternative to Multidimensional Scaling (MDS) that allows to map antipodally invariant distances in the Euclidean space with very little approximation error. The performance of our method, which we named HHC Regression (HHCR), applied to Super-Resolution (SR) improves both in quality (PSNR) and it is faster than any other state-of-the-art method. Additionally, under an application-agnostic interpretation of our regression framework, we also test our algorithm for denoising and depth upscaling with promising results.

Project website (code available)

2015

Naive Bayes Super-Resolution Forest

Jordi Salvador, Eduardo Pérez-Pellitero.

International Conference on Computer Vision (ICCV) 2015, Santiago de Chile.
@inproceedings { Salvador2015,
  author = {Salvador, J. and P\'erez-Pellitero, E.},
  title = {{Naive Bayes Super-Resolution Forest}},
  booktitle = {Proc. {IEEE} Int. Conf. on Computer Vision},
  year = {2015},
}

This paper presents a fast, high-performance method for super resolution with external learning. The first contribution leading to the excellent performance is a bimodal tree for clustering, which successfully exploits the antipodal invariance of the coarse-to-high-res mapping of natural image patches and provides scalability to finer partitions of the underlying coarse patch space. During training an ensemble of such bimodal trees is computed, providing different linearizations of the mapping. The second and main contribution is a fast inference algorithm, which selects the most suitable mapping function within the tree ensemble for each patch by adopting a Local Naive Bayes formulation. The resulting method is beyond one order of magnitude faster and performs objectively and subjectively better than the current state of the art.

Project website (code available)

Accelerating Super-Resolution for 4K Upscaling

Eduardo Pérez-Pellitero, Jordi Salvador, Javier Ruiz-Hidalgo, Bodo Rosenhahn.

Proc. of the International Conference on Consumer Electronics (ICCE) 2015, Las Vegas.
@INPROCEEDINGS{PerezPellitero2015Icce,
  author = {E. Pérez-Pellitero and J. Salvador and J. Ruiz-Hidalgo and B. Rosenhahn},
  title = {Accelerating Super-Resolution for 4K Upscaling},
  booktitle = {IEEE ICCE},
  year = {2015},
}

This paper presents a fast Super-Resolution (SR) algorithm based on a selective patch processing. Motivated by the observation that some regions of images are smooth and unfocused and can be properly upscaled with fast interpolation methods, we locally estimate the probability of performing a degradation-free upscaling. Our proposed framework explores the usage of supervised machine learning techniques and tackles the problem using binary boosted tree classifiers.


2014 Second year of PhD

Fast Super-Resolution via Dense Local Training and Inverse Regressor Search

Eduardo Pérez-Pellitero, Jordi Salvador, Iban Torres, Javier Ruiz-Hidalgo, Bodo Rosenhahn.

Asian Conference on Computer Vision (ACCV) 2014, Singapore.
27% accept rate.
@INPROCEEDINGS{PerezPellitero2014Accv,
  author = {E. Pérez-Pellitero and J. Salvador and I. Torres and Javier Ruiz-Hidalgo and Bodo Rosenhahn},
  title = {Fast Super-Resolution via Dense Local Training and Inverse Regressor Search},
  booktitle = {ACCV},
  year = {2014},
}

Under the locally linear embedding assumption, SR can be properly modeled by a set of linear regressors distributed across the manifold. In this paper we propose a fast inverse-search approach for regression-based SR. Instead of performing a search from the image to the dictionary of regressors, the search is done inversely from the regressors' dictionary to the image patches. Additionally, we propose an improved training scheme for SR linear regressors which improves perceived and objective quality. By merging both contributions we improve both speed and quality compared to the state-of-the-art.


Fast Approximate Nearest-Neighbor Field by Cascaded Spherical Hashing

Iban Torres, Jordi Salvador, Eduardo Pérez-Pellitero.

Asian Conference on Computer Vision (ACCV) 2014, Singapore.
27% accept rate.
@INPROCEEDINGS{Torres2014Accv,
  author = {I. Torres and J. Salvador and E. Pérez-Pellitero },
  title = {Fast Approximate Nearest-Neighbor Field by Cascaded Spherical Hashing},
  booktitle = {ACCV},
  year = {2014},
}

We present an efficient and fast algorithm for computing approximate nearest neighbor fields between two images. Our method builds on the concept of Coherency-Sensitive Hashing (CSH), but uses a recent hashing scheme, Spherical Hashing (SpH), which is known to be better adapted to the nearest-neighbor problem for natural images. Cascaded Spherical Hashing concatenates different configurations of SpH to build larger Hash Tables with less elements in each bin to achieve higher selectivity. Our method amply outperforms existing techniques like PatchMatch and CSH, and the experimental results show that our algorithm is faster and more accurate than existing methods.


Robust Single-Image Super-Resolution using Cross-Scale Self-Similarity

Jordi Salvador, Eduardo Pérez-Pellitero, Axel Torres.
ICIP 2014

@INPROCEEDINGS{Salvador2014Icip,
  author = {Salvador, J. and P\'erez-Pellitero, E. and Kochale, A.},
  title = {{Robust Single-Image Super-Resolution using Cross-Scale Self-Similarity}},
  booktitle = {Proc. IEEE Int. Conf. on Image Processing},
  year = {2014},
}
}

We present a noise-aware single-image super-resolution (SISR) algorithm, which automatically cancels additive noise while adding detail learned from lower-resolution scales. In contrast with most SI-SR techniques, we do not assume the input image to be a clean source of examples. Instead, we adapt the recent and efficient in-place cross-scale self-similarity prior for both learning fine detail examples and reducing image noise. The experimental results show a promising performance, despite the relatively simple algorithm. Both objective evaluations and subjective validations show clear quality improvements when upscaling noisy images.


An Epipolar-Constrained Prior for Efficient Search in Multi-View Scenarios

Ignacio Bosch, Jordi Salvador, Eduardo Pérez-Pellitero, Javier Ruiz-Hidalgo.
EUSIPCO 2014

@INPROCEEDINGS{Bosch2014Eusipco,
  author = {I. Bosch and J. Salvador and E. Pérez-Pellitero and J. Ruiz-Hidalgo},
  title = {An Epipolar-Constrained Prior for Efficient Search in Multi-View Scenarios},
  booktitle = {EUSIPCO},
  year = {2014},
}

In this paper we propose a novel framework for fast exploitation of multi-view cues with applicability in different image processing problems. An epipolar-constrained prior is presented, onto which a random search algorithm is proposed to find good matches among the different views of the same scene. This algorithm includes a generalization of the local coherency in 2D images for multi-view wide-baseline cases. Experimental results show that the geometrical constraint allows a faster initial convergence when finding good matches.


2013 First year of PhD

Bayesian region selection for adaptive dictionary-based Super-Resolution

Eduardo Pérez-Pellitero, Jordi Salvador, Javier Ruiz-Hidalgo, Bodo Rosenhahn.

British Machine Vision Conference (BMVC) 2013, Bristol.
439 submissions
30% accept rate
7% oral accept rate.
@INPROCEEDINGS{PerezPellitero2013Bmvc,
  author = {E. Pérez-Pellitero and J. Salvador and J. Ruiz-Hidalgo and B. Rosenhahn},
  title = {Bayesian region selection for adaptive dictionary-based Super-Resolution},
  booktitle = {BMVC},
  year = {2013},
}

This paper presents a novel sparse SR method, which focuses in adaptively selecting the optimal patches for the dictionary training. The method divides the training images into sub-image regions of sizes that preserve texture consistency. The best-representing region for each input LR patch is found through a Bayesian selection stage. In this selection process, SIFT descriptors are extracted densely from both input LR patches and regions and a local NBNN approach is used in order to efficiently handle the high number of different regions in the training dataset.

Project website

Fast single-image super-resolution with filter selection

Jordi Salvador, Eduardo Pérez-Pellitero, Axel Kochale.
ICIP 2013.

@INPROCEEDINGS{Salvador2013Icip,
  author = {Salvador, J. and Pérez-Pellitero, E. and Kochale, A.},
  title = {Fast single-image super-resolution with filter selection},
  booktitle = {Proc. IEEE Int. Conf. on Image Processing},
  year = {2013},
}

We extend prior work on single-image super-resolution by introducing an adaptive selection of the best fitting upscaling and analysis filters for example learning. This selection is based on local error measurements obtained by using each filter with every image patch, and contrasts with the common approach of a constant metric in both dictionary-based and internal learning super-resolution

Project website

2012 Arriving at Visics in KU Leuven

Object detection using the chains model

E. Pérez-Pellitero
Master Thesis, KU Leuven 2012.

@MASTERSTHESIS{PerezPellitero_MsT_2012,
  author = {E. Pérez-Pellitero and R. Benenson and L. Van Gool},
  title = {Object detection using the chains model},
  school = {KU Leuven},
  year = {2012}
}

This thesis focuses on part-based object detection using as a source of information not only the appearance of the parts themselves, but also the context supplied by other surrounding parts (either inside or outside the object to detect), in opposition to the most used approach of recognizing the object using as a source of information only the appearance of the object itself.


Copyright Notice

The downloadable documents on this web site are presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.

Office N1.012
Empirical Inference Department, Max Planck Institute for Intelligent Systems
Max-Planck-Ring 4
72076 Tübingen
Germany

You can also check my Max Planck website here.

Page created using bootstrap, simplest template. Publication list by R. Benenson.