Occlusion SLAM: Improving Visual SLAM by Leveraging Occluded Points

dc.contributor.authorDhali, Sujan
dc.contributor.authorDasgupta, Bhaskar
dc.date.accessioned2025-07-30T08:08:58Z
dc.date.available2025-07-30T08:08:58Z
dc.date.issued2025
dc.description.abstract-translatedThis paper presents a visual SLAM (simultaneous localization and mapping) system, called Occlusion SLAM, that uses occluded points—map points that are sometimes visible and sometimes blocked by objects in the foreground—and demonstrates a noticeable improvement in camera motion estimation. Occlusion SLAM is built upon ORB SLAM2, with an additional segmentation module based on YOLOv8 and the proposed occluded point detection module. The segmentation module creates a binary mask image of the predefined dynamic objects, which the occlusion detection module uses to identify occluded points, as they are intermittently revealed by the motion of an object. Occlusion SLAM was evaluated on highly dynamic datasets, such as the "walking" and "crowd" sequences from the TUM and Bonn datasets, respectively. Results indicate that the occluded point detection module, combined with segmentation, achieves significant improvements in absolute trajectory estimation with minimal computational requirements compared to ORB SLAM2. The method showed its full potential when at least one-fifth of the map points in a keyframe were identified as occluded points. The experiments also show competitive performance against other well known methods, such as DS SLAM, Dyna SLAM, RDS SLAM, Crowd SLAM and SG SLAM.en
dc.format10 s.cs
dc.format.mimetypeapplication/pdf
dc.identifier.doihttp://www.doi.org/10.24132/JWSCG.2025-8
dc.identifier.issn1213-6972 (print)
dc.identifier.issn1213-6964 (online)
dc.identifier.urihttp://hdl.handle.net/11025/62202
dc.language.isoenen
dc.publisherVáclav Skala - UNION Agencycs
dc.rights© Václav Skala - UNION Agencyen
dc.rights.accessopenAccessen
dc.subjectSimultánní lokalizace a mapování (SLAM)cs
dc.subjectodhad pozicecs
dc.subjectprojekční transformacecs
dc.subjectokluzecs
dc.subjectsémantická segmentacecs
dc.subjectúprava svazkůcs
dc.subject.translatedSimultaneous localization and mapping (SLAM)en
dc.subject.translatedpose estimationen
dc.subject.translatedprojection transformationen
dc.subject.translatedocclusionen
dc.subject.translatedsemantic segmentationen
dc.subject.translatedbundle adjustmenten
dc.titleOcclusion SLAM: Improving Visual SLAM by Leveraging Occluded Pointsen
dc.typečlánekcs
dc.typearticleen
dc.type.statusPeer-revieweden
dc.type.versionpublishedVersionen
local.files.count1*
local.files.size4884505*
local.has.filesyes*

Files

Original bundle
Showing 1 - 1 out of 1 results
No Thumbnail Available
Name:
B83.pdf
Size:
4.66 MB
Format:
Adobe Portable Document Format
License bundle
Showing 1 - 1 out of 1 results
No Thumbnail Available
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description: