Speakers and abstracts

Invited speakers

Extended lectures (1h30)

Talks (30mins)

 

Abstracts

Vision models for emerging technologies in the media industry (EXTENDED LECTURE)

Marcelo Bertalmío (Universitat Pompeu Fabra, Barcelona, Spain)

To enhance the overall viewing experience (for cinema, TV, games, AR/VR) the media industry is continuously striving to improve image quality. Currently the emphasis is on High Dynamic Range (HDR) and Wide Color Gamut (WCG) technologies, which yield images with greater contrast and more vivid colors. The uptake of these technologies, however, has been hampered by the significant open challenges in the science of visual perception. This talk will provide an insight into the science and methods for HDR and WCG, highlighting how the use of vision models is a key element of all state-of-the-art techniques for these emerging technologies, but also how the input from cinema professionals and artists can help vision science in developing, validating and fine-tuning more accurate vision models.

 

Geometric models for color image restoration and their connections to vision science

Thomas Batard (Universitat Pompeu Fabra, Barcelona, Spain) 

In this talk, I present variational models on vector bundles for image restoration which generalize existing Euclidean models. We first establish connections between the proposed models and visual psychophysics and neuroscience. Then, we show that, for well-chosen parameters, the models enable to process simultaneously and independently local (edges, textures) and global (contrast) image features.  

 

Geometric visual hallucinations (EXTENDED LECTURE)

Grégory Faye (IMT, Université Paul Sabatier, Toulouse, France)

In this lecture, I will present basic concepts behind the formation of drug-induced geometric visual hallucinations. The idea is to realize these geometric visual hallucinations as Turing patterns in the primary visual cortex through symmetry breaking bifurcations. In a first time, I will explain the pioneer results of Ermentrout & Cowan when the visual cortex is assumed to be homogeneous, then I will briefly mention the celebrated results of Bressloff et al where the functional architecture of the visual cortex is taken into account by incorporating a feature space (orientation) into the model. Finally, if time allows, I will discuss flicker-induced visual hallucinations. In this later case, the hallucinations are not chemically generated but dynamically induced by the presentation of visual stimuli of flicker type.

 

Visual Illusions Also Deceive Convolutional Neural Networks: Analysis and Implications

Alexander Gomez-Villa (Universitat Pompeu Fabra, Barcelona, Spain)

Visual illusions allow researchers to devise and test new models of visual perception. Here we show that artificial neural networks trained for basic visual tasks in natural images are deceived by brightness and color illusions, having a response that is qualitatively very similar to the human achromatic and chromatic contrast sensitivity functions, and consistent with natural image statistics. We also show that, while these artificial networks are deceived by illusions, their response might be significantly different to that of humans. Our results suggest that low-level illusions appear in any system that has to perform basic visual tasks in natural environments, in line with error minimization explanations of visual function, and they also imply a word of caution on using artificial networks to study human vision, as previously suggested in other contexts in the vision science literature. By the end of the talk we will show how can we use these findings to create new visual illusions.

 

What can resonance frequencies tell us about perception?

Rasa Gulbinaite (Centre de Recherche en Neurosciences de Lyon, France)

The effects of repetitive light stimulation (flicker) on perception and cognition fascinated scientists and artists for almost 200 years. Until very recently, in cognitive neuroscience flicker has been used mainly to “frequency tag” visual stimuli to probe sensory and cognitive processes. This frequency-tagging approach assumes no interaction between the flicker frequency and ongoing endogenous brain rhythms. In contrast, recent evidence suggests that repetitive light stimulation can be used to alter spontaneous brain rhythms and behavior. By using findings from human M/EEG studies and widefield glutamate imaging in mice, I will illustrate how measuring the natural frequencies of the visual system can inform us about its properties and the state.  

 

Cortical models-inspired experiments on active vision : smooth eye movements and the visual processing of structured motion signals

Anna Montagnini (INT, Aix-Marseille Université, France)

The accurate visual tracking of a moving object is a human fundamental skill that allows to reduce the relative slip and instability of the object’s image on the retina, thus granting high-quality vision. In order to optimize tracking performance across time, a quick estimate of the object’s global motion properties needs to be constructed, and dynamically updated, by neuronal populations in the early visual cortices, to be fed to the oculomotor system. However, such estimate can be made difficult by different sources of uncertainty (e.g. the complex and noisy spatiotemporal structure of the stimuli) and ambiguity (e.g. the “aperture problem”). Concurrently, tracking performance can be greatly improved in terms of latency and accuracy by taking into account predictive cues, especially under variable conditions of visibility and in presence of ambiguous retinal information.I will review several recent studies focusing on the integration of retinal and extra-retinal information for the control of human smooth pursuit. By dynamically probing the motion tracking performance with oculomotor and perceptual experimental paradigms, we provide a benchmark to test models of cortical motion processing and general theoretical hypotheses about probabilistic inference in the brain. 

 

A kernel-driven model of lateral connections in V1 

Noemi Montobbio (Italian Institute of Technology, Genoa, Italy)

The lateral connectivity of the primary visual cortex (V1) is known to develop in an orientation specific fashion, and it is believed to play an important role in the ability of our visual system to perform contour integration. We developed a metric model to describe the local geometry of such connections, through a kernel defined by the pairwise correlations between receptive profiles of simple cells. We modeled the spreading of neural activity across the cortex via a kernel-driven diffusion in the cortical space endowed with the metric induced by the receptive profiles. The patterns obtained prove compatible with both psychophysical and neurophysiological data. We then employed this approach to insert parameter-free associative lateral connections in Convolutional Neural Networks (CNNs), to improve their robustness to image perturbations such as occlusions. 

 

The implication of V1 horizontal connectivity in the apparent motion and flash-lag effect 

Cyril Monier (Université Paris-Saclay, CNRS, Institut des Neurosciences Paris-Saclay, 91190, Gif-sur-Yvette, France.)

TBA

 

 

Neural field model of V1 orientation selectivity to reconcile structure and function

James Rankin (University of Exeter, UK) 

Voltage-sensitive dye imaging experiments in primary visual cortex (V1) have shown that local, oriented visual stimuli elicit stable orientation-selective activation within the stimulus retinotopic footprint (Chavane et al 2011). The cortical activation dynamically extends far beyond the retinotopic footprint, but the peripheral spread stays non-selective - a surprising finding given a number of anatomo-functional studies showing the orientation specificity of long-range connections. Here we use a computational model to investigate this apparent discrepancy by studying the expected population response using known published anatomical constraints. The dynamics of input-driven localized states were simulated in a planar neural field model with multiple sub-populations encoding orientation. The realistic connectivity profile has parameters controlling the clustering of long-range connections and their orientation bias. We found substantial overlap between the anatomically relevant parameter range and a steep decay in orientation selective activation that is consistent with the imaging experiments.  In this way our study reconciles the reported orientation bias of long-range connections with the functional expression of orientation selective neural activity. Our results demonstrate this sharp decay is contingent on three factors, that long-range connections are sufficiently diffuse, that the orientation bias of these connections is in an intermediate range (consistent with anatomy) and that excitation is sufficiently balanced by inhibition. Conversely, our modelling results predict that, for reduced global inhibition strength, spurious orientation selective activation could be generated through long-range lateral connections. Furthermore, if the orientation bias of lateral connections is very strong, or if inhibition is particularly weak, the network operates close to an instability leading to unbounded cortical activation. 

 

The differential brain: from neurogeometry to heterogenesis (EXTENDED LECTURE)

 

Alessandro Sarti (CNRS, EHESS)

We will briefly review some basic concept of neurogeometry to model the functional architecture of the primary visual cortex. While classic neurogeometrical models deal with an homogenous class of cells, we will problematize the heterogeneity of connectivity geometry and dynamics

 

Visual hallucinations in a neural field model for color perception unifying assimilation and contrast

Romain Veltz (MathNeuro Team, Inria Sophia Antipolis)

We study the visual hallucinations which can be produced by a recent model of color perception unifying assimilation and contrast. This model, which relies on the notion of color opponency introduced by Hering, has been previously tuned [Song et al. 2019] to reproduce some nontrivial behaviors of the color shifts observed in experiments. Here, we perform bifurcation analysis, based on the properties of Wiener-Hopf operators, of this planar model to predict visual hallucinations. Numerical simulations are provided to assess the global stability of the predicted visual hallucinations. 

Online user: 1