Adaptive Motion Pooling and Diffusion for Optical Flow Computation - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2017

Adaptive Motion Pooling and Diffusion for Optical Flow Computation

Abstract

We propose to extend a state of the art bio-inspired model for optic flow computation through adaptive processing by focusing on the role of local context indicative of the local velocity estimates reliability. We set a network structure representative of cortical areas V1, V2 and MT, and incorporate three functional principles observed in primate visual system: contrast adaptation, adaptive afferent pooling and MT diffusion that are adaptive dependent upon the 2D image structure (Adaptive Motion Pooling and Diffusion, AMPD). We assess the AMPD performance on Middlebury optical flow estimation dataset, showing that the proposed AMPD model performs better than the baseline one and its overall performance is comparable with many computer vision methods.
Fichier principal
Vignette du fichier
paper_wbicv.pdf (4.27 Mo) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01589983 , version 1 (19-09-2017)

Identifiers

  • HAL Id : hal-01589983 , version 1

Cite

N S Kartheek Medathati, Manuela S Chessa, Guillaume S Masson, Pierre Kornprobst, Fabio S Solari. Adaptive Motion Pooling and Diffusion for Optical Flow Computation. WBICV 2017 : First International Workshop on Brain-Inspired Computer Vision, Sep 2017, Catania, Sicily, Italy. ⟨hal-01589983⟩
253 View
203 Download

Share

Gmail Facebook X LinkedIn More