arXiv subjects

Jean-Philippe Thiran

Publications and source records attributed to Jean-Philippe Thiran.

90 records · Page 5Linked to original sources

arXiv2018

Robust Real-Time Multi-View Eye Tracking

Despite significant advances in improving the gaze tracking accuracy under controlled conditions, the tracking robustness under real-world conditions, such as large head pose and movements, use of eyeglasses, illumination and eye type variations, remains a major challenge in eye tracking. In this paper, we revisit this challenge and introduce a real-time multi-camera eye tracking framework to improve the tracking robustness. First, differently from previous work, we design a multi-view tracking setup that allows for acquiring multiple eye appearances simultaneously. Leveraging multi-view appearances enables to more reliably detect gaze features under challenging conditions, particularly when they are obstructed in conventional single-view appearance due to large head movements or eyewear effects. The features extracted on various appearances are then used for estimating multiple gaze outputs. Second, we propose to combine estimated gaze outputs through an adaptive fusion mechanism to compute user's overall point of regard. The proposed mechanism firstly determines the estimation reliability of each gaze output according to user's momentary head pose and predicted gazing behavior, and then performs a reliability-based weighted fusion. We demonstrate the efficacy of our framework with extensive simulations and user experiments on a collected dataset featuring 20 subjects. Our results show that in comparison with state-of-the-art eye trackers, the proposed framework provides not only a significant enhancement in accuracy but also a notable robustness. Our prototype system runs at 30 frames-per-second (fps) and achieves 1 degree accuracy under challenging experimental scenarios, which makes it suitable for applications demanding high accuracy and robustness.

cs.CV↗

arXiv2017

A Computer Vision System to Localize and Classify Wastes on the Streets

Littering quantification is an important step for improving cleanliness of cities. When human interpretation is too cumbersome or in some cases impossible, an objective index of cleanliness could reduce the littering by awareness actions. In this paper, we present a fully automated computer vision application for littering quantification based on images taken from the streets and sidewalks. We have employed a deep learning based framework to localize and classify different types of wastes. Since there was no waste dataset available, we built our acquisition system mounted on a vehicle. Collected images containing different types of wastes. These images are then annotated for training and benchmarking the developed system. Our results on real case scenarios show accurate detection of littering on variant backgrounds.

cs.CV↗

arXiv2017

Visual Speech Recognition Using PCA Networks and LSTMs in a Tandem GMM-HMM System

Automatic visual speech recognition is an interesting problem in pattern recognition especially when audio data is noisy or not readily available. It is also a very challenging task mainly because of the lower amount of information in the visual articulations compared to the audible utterance. In this work, principle component analysis is applied to the image patches - extracted from the video data - to learn the weights of a two-stage convolutional network. Block histograms are then extracted as the unsupervised learning features. These features are employed to learn a recurrent neural network with a set of long short-term memory cells to obtain spatiotemporal features. Finally, the obtained features are used in a tandem GMM-HMM system for speech recognition. Our results show that the proposed method has outperformed the baseline techniques applied to the OuluVS2 audiovisual database for phrase recognition with the frontal view cross-validation and testing sentence correctness reaching 79% and 73%, respectively, as compared to the baseline of 74% on cross-validation.

cs.CV↗

arXiv2017

Combining LiDAR Space Clustering and Convolutional Neural Networks for Pedestrian Detection

Pedestrian detection is an important component for safety of autonomous vehicles, as well as for traffic and street surveillance. There are extensive benchmarks on this topic and it has been shown to be a challenging problem when applied on real use-case scenarios. In purely image-based pedestrian detection approaches, the state-of-the-art results have been achieved with convolutional neural networks (CNN) and surprisingly few detection frameworks have been built upon multi-cue approaches. In this work, we develop a new pedestrian detector for autonomous vehicles that exploits LiDAR data, in addition to visual information. In the proposed approach, LiDAR data is utilized to generate region proposals by processing the three dimensional point cloud that it provides. These candidate regions are then further processed by a state-of-the-art CNN classifier that we have fine-tuned for pedestrian detection. We have extensively evaluated the proposed detection process on the KITTI dataset. The experimental results show that the proposed LiDAR space clustering approach provides a very efficient way of generating region proposals leading to higher recall rates and fewer misses for pedestrian detection. This indicates that LiDAR data can provide auxiliary information for CNN-based approaches.

cs.CV↗

arXiv2017

Robust dimensionality reduction for interferometric imaging of Cygnus A

Extremely high data rates expected in next-generation radio interferometers necessitate a fast and robust way to process measurements in a big data context. Dimensionality reduction can alleviate computational load needed to process these data, in terms of both computing speed and memory usage. In this article, we present image reconstruction results from highly reduced radio-interferometric data, following our previously proposed data dimensionality reduction method, $\mathrm{R}_{\mathrm{sing}}$, based on studying the distribution of the singular values of the measurement operator. This method comprises a simple weighted, subsampled discrete Fourier transform of the dirty image. Additionally, we show that an alternative gridding-based reduction method works well for target data sizes of the same order as the image size. We reconstruct images from well-calibrated VLA data to showcase the robustness of our proposed method down to very low data sizes in a 'real data' setting. We show through comparisons with the conventional reduction method of time- and frequency-averaging, that our proposed method produces more accurate reconstructions while reducing data size much further, and is particularly robust when data sizes are aggressively reduced to low fractions of the image size. $\mathrm{R}_{\mathrm{sing}}$ can function in a block-wise fashion, and could be used in the future to process incoming data by blocks in real-time, thus opening up the possibility of performing 'on-line' imaging as the data are being acquired. MATLAB code for the proposed dimensionality reduction method is available on GitHub.

astro-ph.IM↗

arXiv2017

A Fourier dimensionality reduction model for big data interferometric imaging

Data dimensionality reduction in radio interferometry can provide savings of computational resources for image reconstruction through reduced memory footprints and lighter computations per iteration, which is important for the scalability of imaging methods to the big data setting of the next-generation telescopes. This article sheds new light on dimensionality reduction from the perspective of compressed sensing theory and studies its interplay with imaging algorithms designed in the context of convex optimization. We propose a post-gridding linear data embedding to the space spanned by the left singular vectors of the measurement operator, providing a dimensionality reduction below image size. This embedding preserves the null space of the measurement operator and hence also its sampling properties as per compressed sensing theory. We show that this can be approximated by first computing the dirty image and then applying a weighted subsampled discrete Fourier transform to obtain the final reduced data vector. This Fourier dimensionality reduction model ensures a fast implementation of the full measurement operator, essential for any iterative image reconstruction method. The proposed reduction also preserves the i.i.d. Gaussian properties of the original measurement noise. For convex optimization-based imaging algorithms, this is key to justify the use of the standard L2-norm as the data fidelity term. Our simulations confirm that this dimensionality reduction approach can be leveraged by convex optimization algorithms with no loss in imaging quality relative to reconstructing the image from the complete visibility data set. Reconstruction results in simulation settings with no direction dependent effects or calibration errors show promising performance of the proposed dimensionality reduction. MATLAB code implementing the proposed reduction method is available on GitHub.

astro-ph.IM↗

arXiv2016

Scalable splitting algorithms for big-data interferometric imaging in the SKA era

In the context of next generation radio telescopes, like the Square Kilometre Array, the efficient processing of large-scale datasets is extremely important. Convex optimisation tasks under the compressive sensing framework have recently emerged and provide both enhanced image reconstruction quality and scalability to increasingly larger data sets. We focus herein mainly on scalability and propose two new convex optimisation algorithmic structures able to solve the convex optimisation tasks arising in radio-interferometric imaging. They rely on proximal splitting and forward-backward iterations and can be seen, by analogy with the CLEAN major-minor cycle, as running sophisticated CLEAN-like iterations in parallel in multiple data, prior, and image spaces. Both methods support any convex regularisation function, in particular the well studied l1 priors promoting image sparsity in an adequate domain. Tailored for big-data, they employ parallel and distributed computations to achieve scalability, in terms of memory and computational requirements. One of them also exploits randomisation, over data blocks at each iteration, offering further flexibility. We present simulation results showing the feasibility of the proposed methods as well as their advantages compared to state-of-the-art algorithmic solvers. Our Matlab code is available online on GitHub.

astro-ph.IM↗

arXiv2016

A comprehensive error rate for multiple testing

The higher criticism of a family of tests starts with the individual uncorrected p-values of each test. It then requires a procedure for deciding whether the collection of p-values indicates the presence of a real effect and if possible selects the ones that deserve closer scrutiny. This paper investigates procedures in which the ordered p-values are compared to an arbitrary positive and non-decreasing threshold sequence.

math.ST↗

arXiv2015

Spherical deconvolution of multichannel diffusion MRI data with non-Gaussian noise models and spatial regularization

Spherical deconvolution (SD) methods are widely used to estimate the intra-voxel white-matter fiber orientations from diffusion MRI data. However, while some of these methods assume a zero-mean Gaussian distribution for the underlying noise, its real distribution is known to be non-Gaussian and to depend on the methodology used to combine multichannel signals. Indeed, the two prevailing methods for multichannel signal combination lead to Rician and noncentral Chi noise distributions. Here we develop a Robust and Unbiased Model-BAsed Spherical Deconvolution (RUMBA-SD) technique, intended to deal with realistic MRI noise, based on a Richardson-Lucy (RL) algorithm adapted to Rician and noncentral Chi likelihood models. To quantify the benefits of using proper noise models, RUMBA-SD was compared with dRL-SD, a well-established method based on the RL algorithm for Gaussian noise. Another aim of the study was to quantify the impact of including a total variation (TV) spatial regularization term in the estimation framework. To do this, we developed TV spatially-regularized versions of both RUMBA-SD and dRL-SD algorithms. The evaluation was performed by comparing various quality metrics on 132 three-dimensional synthetic phantoms involving different inter-fiber angles and volume fractions, which were contaminated with noise mimicking patterns generated by data processing in multichannel scanners. The results demonstrate that the inclusion of proper likelihood models leads to an increased ability to resolve fiber crossings with smaller inter-fiber angles and to better detect non-dominant fibers. The inclusion of TV regularization dramatically improved the resolution power of both techniques. The above findings were also verified in brain data.

physics.med-ph↗

arXiv2015

Generative models of the human connectome

The human connectome represents a network map of the brain's wiring diagram and the pattern into which its connections are organized is thought to play an important role in cognitive function. The generative rules that shape the topology of the human connectome remain incompletely understood. Earlier work in model organisms has suggested that wiring rules based on geometric relationships (distance) can account for many but likely not all topological features. Here we systematically explore a family of generative models of the human connectome that yield synthetic networks designed according to different wiring rules combining geometric and a broad range of topological factors. We find that a combination of geometric constraints with a homophilic attachment mechanism can create synthetic networks that closely match many topological characteristics of individual human connectomes, including features that were not included in the optimization of the generative model itself. We use these models to investigate a lifespan dataset and show that, with age, the model parameters undergo progressive changes, suggesting a rebalancing of the generative factors underlying the connectome across the lifespan.

q-bio.NC↗

arXiv2014

MBIS: Multivariate Bayesian Image Segmentation Tool

We present MBIS (Multivariate Bayesian Image Segmentation tool), a clustering tool based on the mixture of multivariate normal distributions model. MBIS supports multi-channel bias field correction based on a B-spline model. A second methodological novelty is the inclusion of graph-cuts optimization for the stationary anisotropic hidden Markov random field model. Along with MBIS, we release an evaluation framework that contains three different experiments on multi-site data. We first validate the accuracy of segmentation and the estimated bias field for each channel. MBIS outperforms a widely used segmentation tool in a cross-comparison evaluation. The second experiment demonstrates the robustness of results on atlas-free segmentation of two image sets from scan-rescan protocols on 21 healthy subjects. Multivariate segmentation is more replicable than the monospectral counterpart on T1-weighted images. Finally, we provide a third experiment to illustrate how MBIS can be used in a large-scale study of tissue volume change with increasing age in 584 healthy subjects. This last result is meaningful as multivariate segmentation performs robustly without the need for prior knowledge

cs.CV↗

arXiv2013

Sparse regularization for fiber ODF reconstruction: from the suboptimality of $\ell_2$ and $\ell_1$ priors to $\ell_0$

Diffusion MRI is a well established imaging modality providing a powerful way to probe the structure of the white matter non-invasively. Despite its potential, the intrinsic long scan times of these sequences have hampered their use in clinical practice. For this reason, a large variety of methods have been recently proposed to shorten the acquisition times. Among them, spherical deconvolution approaches have gained a lot of interest for their ability to reliably recover the intra-voxel fiber configuration with a relatively small number of data samples. To overcome the intrinsic instabilities of deconvolution, these methods use regularization schemes generally based on the assumption that the fiber orientation distribution (FOD) to be recovered in each voxel is sparse. The well known Constrained Spherical Deconvolution (CSD) approach resorts to Tikhonov regularization, based on an l2-norm prior, which promotes a weak version of sparsity. Also, in the last few years compressed sensing has been advocated to further accelerate the acquisitions and l1-norm minimization is generally employed as a means to promote sparsity in the recovered FODs. In this paper, we provide evidence that the use of an l1-norm prior to regularize this class of problems is somewhat inconsistent with the fact that the fiber compartments all sum up to unity. To overcome this l1 inconsistency while simultaneously exploiting sparsity more optimally than through an l2 prior, we reformulate the reconstruction problem as a constrained formulation between a data term and and a sparsity prior consisting in an explicit bound on the l0 norm of the FOD, i.e. on the number of fibers. The method has been tested both on synthetic and real data. Experimental results show that the proposed l0 formulation significantly reduces modeling errors compared to the state-of-the-art l2 and l1 regularization approaches.

q-bio.QM↗

arXiv2013

Tensor optimisation for optical-interferometric imaging

Image recovery in optical interferometry is an ill-posed nonlinear inverse problem arising from incomplete power spectrum and bispectrum measurements. We reformulate this nonlin- ear problem as a linear problem for the supersymmetric rank-1 order-3 tensor formed by the tensor product of the vector representing the image under scrutiny with itself. On one hand, we propose a linear convex approach for tensor recovery with built-in supersymmetry, and regularising the inverse problem through a nuclear norm relaxation of a low-rank constraint. On the other hand, we also study a nonlinear nonconvex approach with built-in rank-1 con- straint but where supersymmetry is relaxed, formulating the problem for the tensor product of 3 vectors. In this second approach, only linear convex minimisation subproblems are how- ever solved, alternately and iteratively for the 3 vectors. We provide a comparative analysis of these two novel approaches through numerical simulations on small-size images.

astro-ph.IM↗

arXiv2013

Two step multiple comparisons procedures for positively dependent data with application to detecting differences in human brain network topologies

We consider the problem of testing positively dependent multiple hypotheses assuming that a prior information about the dependence structure is available. We propose two-step multiple comparisons procedures that exploit the prior information of the dependence structure, without relying on strong assumptions. In the first step, we group the tests into subsets where tests are supposed to be positively dependent and in each of which we compute the standardized mean of the test scores. Given the subset mean scores or equivalently the subsets p-values, we apply a first screening at a predefined threshold, which results in two types of subsets. Based on this typing, the original single test p-values are modified such that they can be used in conjunction with any multiple comparison procedure. We show by means of different simulation that power is gained with the proposed two-step methods, and compare it with traditional multiple comparison procedures. As an illustration, our method is applied on real data comparing topological differences between two groups of human brain networks.

stat.ME↗

arXiv2013

Optimality in multiple comparison procedures

When many (m) null hypotheses are tested with a single dataset, the control of the number of false rejections is often the principal consideration. Two popular controlling rates are the probability of making at least one false discovery (FWER) and the expected fraction of false discoveries among all rejections (FDR). Scaled multiple comparison error rates form a new family that bridges the gap between these two extremes. For example, the Scaled Expected Value (SEV) limits the number of false positives relative to an arbitrary increasing function of the number of rejections, that is, E(FP/s(R)). We discuss the problem of how to choose in practice which procedure to use, with elements of an optimality theory, by considering the number of false rejections FP separately from the number of correct rejections TP. Using this framework we will show how to choose an element in the new family mentioned above.

stat.ME↗

arXiv2013

Sparsity Averaging for Compressive Imaging

We discuss a novel sparsity prior for compressive imaging in the context of the theory of compressed sensing with coherent redundant dictionaries, based on the observation that natural images exhibit strong average sparsity over multiple coherent frames. We test our prior and the associated algorithm, based on an analysis reweighted $\ell_1$ formulation, through extensive numerical simulations on natural images for spread spectrum and random Gaussian acquisition schemes. Our results show that average sparsity outperforms state-of-the-art priors that promote sparsity in a single orthonormal basis or redundant frame, or that promote gradient sparsity. Code and test data are available at https://github.com/basp-group/sopt.

cs.IT↗

arXiv2012

Enhanced Compressed Sensing Recovery with Level Set Normals

We propose a compressive sensing algorithm that exploits geometric properties of images to recover images of high quality from few measurements. The image reconstruction is done by iterating the two following steps: 1) estimation of normal vectors of the image level curves and 2) reconstruction of an image fitting the normal vectors, the compressed sensing measurements and the sparsity constraint. The proposed technique can naturally extend to non local operators and graphs to exploit the repetitive nature of textured images in order to recover fine detail structures. In both cases, the problem is reduced to a series of convex minimization problems that can be efficiently solved with a combination of variable splitting and augmented Lagrangian methods, leading to fast and easy-to-code algorithms. Extended experiments show a clear improvement over related state-of-the-art algorithms in the quality of the reconstructed images and the robustness of the proposed method to noise, different kind of images and reduced measurements.

cs.CV↗

arXiv2012

Spread spectrum magnetic resonance imaging

We propose a novel compressed sensing technique to accelerate the magnetic resonance imaging (MRI) acquisition process. The method, coined spread spectrum MRI or simply s2MRI, consists of pre-modulating the signal of interest by a linear chirp before random k-space under-sampling, and then reconstructing the signal with non-linear algorithms that promote sparsity. The effectiveness of the procedure is theoretically underpinned by the optimization of the coherence between the sparsity and sensing bases. The proposed technique is thoroughly studied by means of numerical simulations, as well as phantom and in vivo experiments on a 7T scanner. Our results suggest that s2MRI performs better than state-of-the-art variable density k-space under-sampling approaches

cs.OH↗