Motion Estimation and Segmentation of Natural Phenomena

Research output: Contribution to conferencePaperpeer-review

151 Downloads (Pure)

Abstract

Dense motion estimation for dynamic natural phenomena (water, smoke, fire, etc.) is a significant open problem. Current approaches tend to be either general, giving poor results, or specialise in one phenomenon and fail to generalise. Segmentation of phenomena is also an open problem. This paper describes an approach to estimate dense motion for dynamic phenomena that is simple, general, and which yields state of the art results. We use our dense motion field to segment phenomena to above state of the art levels. We demonstrate our contributions using lab-based video, video from a public dataset, and from the internet.
Original languageEnglish
Publication statusPublished - 4 Sep 2018
Event29th British Machine Vision Conference, BMVC 2018 - Northumbria University, Newcastle, UK United Kingdom
Duration: 3 Sep 20186 Sep 2018
http://bmvc2018.org/

Conference

Conference29th British Machine Vision Conference, BMVC 2018
Abbreviated titleBMVC 2018
Country/TerritoryUK United Kingdom
CityNewcastle
Period3/09/186/09/18
Internet address

Keywords

  • natutal phenomena, tracking, segmentation

Fingerprint

Dive into the research topics of 'Motion Estimation and Segmentation of Natural Phenomena'. Together they form a unique fingerprint.

Cite this