arXiv subjects

Nicolas Guérin

Publications and source records attributed to Nicolas Guérin.

4 recordsLinked to original sources

arXiv2022

Investigating the Role of Image Retrieval for Visual Localization -- An exhaustive benchmark

Visual localization, i.e., camera pose estimation in a known scene, is a core component of technologies such as autonomous driving and augmented reality. State-of-the-art localization approaches often rely on image retrieval techniques for one of two purposes: (1) provide an approximate pose estimate or (2) determine which parts of the scene are potentially visible in a given query image. It is common practice to use state-of-the-art image retrieval algorithms for both of them. These algorithms are often trained for the goal of retrieving the same landmark under a large range of viewpoint changes which often differs from the requirements of visual localization. In order to investigate the consequences for visual localization, this paper focuses on understanding the role of image retrieval for multiple visual localization paradigms. First, we introduce a novel benchmark setup and compare state-of-the-art retrieval representations on multiple datasets using localization performance as metric. Second, we investigate several definitions of "ground truth" for image retrieval. Using these definitions as upper bounds for the visual localization paradigms, we show that there is still sgnificant room for improvement. Third, using these tools and in-depth analysis, we show that retrieval performance on classical landmark retrieval or place recognition tasks correlates only for some but not all paradigms to localization performance. Finally, we analyze the effects of blur and dynamic scenes in the images. We conclude that there is a need for retrieval approaches specifically designed for localization paradigms. Our benchmark and evaluation protocols are available at https://github.com/naver/kapture-localization.

cs.CV↗

arXiv2021

Large-scale Localization Datasets in Crowded Indoor Spaces

Estimating the precise location of a camera using visual localization enables interesting applications such as augmented reality or robot navigation. This is particularly useful in indoor environments where other localization technologies, such as GNSS, fail. Indoor spaces impose interesting challenges on visual localization algorithms: occlusions due to people, textureless surfaces, large viewpoint changes, low light, repetitive textures, etc. Existing indoor datasets are either comparably small or do only cover a subset of the mentioned challenges. In this paper, we introduce 5 new indoor datasets for visual localization in challenging real-world environments. They were captured in a large shopping mall and a large metro station in Seoul, South Korea, using a dedicated mapping platform consisting of 10 cameras and 2 laser scanners. In order to obtain accurate ground truth camera poses, we developed a robust LiDAR SLAM which provides initial poses that are then refined using a novel structure-from-motion based optimization. We present a benchmark of modern visual localization algorithms on these challenging datasets showing superior performance of structure-based methods using robust image features. The datasets are available at: https://naverlabs.com/datasets

cs.CV↗

arXiv2017

Homogenization and Clustering as a Non-Statistical Methodology to Assess Multi-Parametrical Chain Problems

We present a new theoretical and numerical assessment methodology for a one-dimensional process chain with general applicability to management problems such as the optimization of decision chains or production chains. The process is thereby seen as a chain of subsequently arranged units with random parameters influencing the objective function. For solving such complex chain problems, analytical methods usually fail and statistical methods only provide approximate solutions while requiring massive computing power. We took insights from physics to develop a new methodology based on homogenization and clustering. The core idea is to replace the complex real chain with a virtual chain that homogenizes the involved parameters and clusters the working units into global units to facilitate computation. This methodology drastically reduces computing time, allows for the derivation of analytical formulas, and provides fast and objective insights about the optimization problem under investigation. We illustrate the analytical potency of this methodology by applying it to the production problem of selecting the economically superior quality maintenance strategy. It can further be applied to all sequential multi-parametrical chain problems commonly found in business.

econ.GN↗

arXiv2015

Symmetry of the Fermi surface and evolution of the electronic structure across the paramagnetic-helimagnetic transition in MnSi/Si(111)

MnSi has been extensively studied for five decades, nonetheless detailed information on the Fermi surface (FS) symmetry is still lacking. This missed information prevented from a comprehensive understanding the nature of the magnetic interaction in this material. Here, by performing angle-resolved photoemission spectroscopy on high-quality MnSi films epitaxially grown on Si(111), we unveil the FS symmetry and the evolution of the electronic structure across the paramagnetic-helimagnetic transition at T$_C$ $\sim$ 40 K, along with the appearance of sharp quasiparticle emission below T$_C$. The shape of the resulting FS is found to fulfill robust nesting effects. These effects can be at the origin of strong magnetic fluctuations not accounted for by state-of-art quasiparticle self-consistent GW approximation. From this perspective, the unforeseen quasiparticle damping detected in the paramagnetic phase and relaxing only below T$_C$, along with the persistence of the d-bands splitting well above T$_C$, at odds with a simple Stoner model for itinerant magnetism, open the search for exotic magnetic interactions favored by FS nesting and affecting the quasiparticles lifetime.

cond-mat.str-el↗