Towards a bio-inspired evaluation methodology for motion estimation models
نویسندگان
چکیده
Offering proper evaluation methodology is essential to continue progress in modelling neural mechanisms in vision/visual information processing. Currently, evaluation of motion estimation models lacks a proper methodology for comparing their performance against the visual system. Here, we set the basis for such a new benchmark methodology which is based on human visual performance as measured in psychophysics, ocular following and neurobiology. This benchmark will enable comparisons between different kinds of models, but also it will challenge current motion estimation models and better characterize their properties with respect to visual cortex performance. To do so, we propose a database of image sequences taken from neuroscience and psychophysics literature. In this article, we focus on two aspects of motion estimation, which are the dynamics of motion integration and the respective influence between 1D versus 2D cues. Then, since motion models possibly deal with different kinds of motion representations and scale, we define here two general readouts based on a global motion estimation. Such readouts, namely eye movements and perceived motion will serve as a reference to compare simulated and experimental data. We evaluate the performance of several models on this data to establish the current state of the art. Models chosen for comparison have very different properties and internal mechanisms, such as feedforward normalisation of V1 and MT processing and recurrent feedback. As a whole, we provide here the basis for a valuable evaluation methodology to unravel the fundamental mechanisms of the visual cortex in motion perception. Our database is freely available on the web together with scoring instructions and results at http://www-sop.inria.fr/neuromathcomp/software/motionpsychobench. Key-words: neural models; smooth pursuit; motion perception; benchmark ∗ [email protected] † [email protected] ‡ [email protected] § [email protected] ¶ [email protected] in ria -0 04 92 00 1, v er si on 1 14 J un 2 01 0 Vers une méthodologie bio-inspirée d’évaluation de modèles d’estimation du mouvement Résumé : Offrir une méthodologie d’évaluation est essentiel pour la recherche en modélisation des mécanismes neuraux impliqués dans la vision. Actuellement, il manque à l’évaluation des modèles d’estimation du mouvement une méthodologie bien définie permettant de comparer leurs performances avec celles du système visuel. Ici nous posons les bases d’un tel banc d’essai, basé sur les performances visuelles des humains telles que mesurées en psychophysique, en oculo-motricité, et en neurobiologie. Ce banc d’essai permettra de comparer différents modèles, mais aussi de mieux caractériser leurs propriétés en regard du comportement du système visuel. Dans ce but, nous proposons un ensemble de séquences vidéos, issues des expérimentations en neurosciences et en psychophysique. Dans cet article, nous mettons l’accent sur deux principaux aspects de l’estimation du mouvement : les dynamiques d’intégration du mouvement, et les influences respectives des informations 1D par rapport aux informations 2D. De là, nous définissons deux lectures basés sur l’estimation du mouvement global. De telles lectures , nommément les mouvements des yeux, et le mouvement perçu, serviront de référence pour comparer les données expérimentales et simulées. Nous évaluons les performances de différents modèles sur ces stimuli afin d’établir un état de l’art des modèles d’intégration du mouvement. Les modèles comparés sont choisis en fonction de leurs grandes différences en terme de propriétes et de mécanismes internes (rétroaction, normalisation). En définitive, nous établissons dans ce travail les bases d’une méthodologie d’évaluation permettant de découvrir les mécanismes fondamentaux du cortex visuel dédié à la perception du mouvement. Notre jeu de stimuli est librement accessible sur Internet, accompagné d’instructions pour l’évaluation, et de résultats, à l’adresse : http://www-sop.inria.fr/neuromathcomp/software/ motionpsychobench. Mots-clés : modèles neuraux; poursuite lisse; perception du mouvement; banc d’essai in ria -0 04 92 00 1, v er si on 1 14 J un 2 01 0 A bio-inspired evaluation methodology for motion estimation 3
منابع مشابه
Evaluating Motion Estimation Models from Behavioural and Psychophysical Data
Offering proper evaluation methodology is essential to continue progress in modelling the neural mechanisms involved in vision information processing. Currently the evaluation of biologically inspired motion estimation models lacks a proper methodology for comparing their performance against behavioural and psychophysical data. Here we set the basis for such a new benchmark methodology based on...
متن کاملOn-chip visual perception of motion: A bio-inspired connectionist model on FPGA
Visual motion provides useful information to understand the dynamics of a scene to allow intelligent systems interact with their environment. Motion computation is usually restricted by real time requirements that need the design and implementation of specific hardware architectures. In this paper, the design of hardware architecture for a bio-inspired neural model for motion estimation is pres...
متن کاملBio-inspired Speed Detection and Discrimination
In the field of computer vision, a crucial task is the detection of motion (also called optical flow extraction). This operation allows analysis such as 3D reconstruction, feature tracking, time-to-collision and novelty detection among others. Most of the optical flow extraction techniques work within a finite range of speeds. Usually, the range of detection is extended towards higher speeds by...
متن کاملA Fast Adaptive Block-matching Motion Estimation Algorithm
In this work, we have developed a new bio-inspired neural network algorithms for blockbased motion estimation. The main goal is to bridge the gap between algorithmic and biological vision by suggesting a bio-inspired motion estimation model based on neural network. We simplify the matching criterion for the block matching algorithm to reduce the hardware complexity and a number of input ports w...
متن کاملOPTIMUM PERFORMANCE-BASED DESIGN OF CONCENTRICALLY BRACED STEEL FRAMES SUBJECTED TO NEAR-FAULT GROUND MOTION EXCITATIONS
This paper presents a practical methodology for optimization of concentrically braced steel frames subjected to forward directivity near-fault ground motions, based on the concept of uniform deformation theory. This is performed by gradually shifting inefficient material from strong parts of the structure to the weak areas until a state of uniform deformation is achieved. In this regard, to ove...
متن کامل