publication . Article . 2020

Implicit and Explicit Regularization for Optical Flow Estimation

Konstantinos Karageorgos; Anastasios Dimou; Federico Alvarez; Petros Daras;
Open Access
  • Published: 01 Jul 2020 Journal: Sensors, volume 20, page 3,855 (eissn: 1424-8220, Copyright policy)
  • Publisher: MDPI AG
In this paper, two novel and practical regularizing methods are proposed to improve existing neural network architectures for monocular optical flow estimation. The proposed methods aim to alleviate deficiencies of current methods, such as flow leakage across objects and motion consistency within rigid objects, by exploiting contextual information. More specifically, the first regularization method utilizes semantic information during the training process to explicitly regularize the produced optical flow field. The novelty of this method lies in the use of semantic segmentation masks to teach the network to implicitly identify the semantic edges of an object an...
Persistent Identifiers
free text keywords: Electrical and Electronic Engineering, Analytical Chemistry, Atomic and Molecular Physics, and Optics, Biochemistry, Article, optical flow, regularization, semantic segmentation, motion consistency, coordconv, Computer science, Artificial neural network, Spatial contextual awareness, Pixel, Regularization (mathematics), Segmentation, Convergence (routing), Optical flow, Inference, Algorithm, lcsh:Chemical technology, lcsh:TP1-1185
Funded by
Advanced tools for fighting oNline Illegal TrAfficking
  • Funder: European Commission (EC)
  • Project Code: 787061
  • Funding stream: H2020 | RIA
Validated by funder
Download fromView all 5 versions
Open Access
Provider: Sygma
Open Access
Article . 2020
Provider: ZENODO
Open Access
Article . 2020
Any information missing or wrong?Report an Issue