Show simple item record

dc.contributor.advisorWonka, Peter
dc.contributor.authorBhat, Shariq
dc.date.accessioned2021-04-22T07:58:31Z
dc.date.available2021-04-22T07:58:31Z
dc.date.issued2021-04-21
dc.identifier.citationBhat, S. (2021). Depth Estimation Using Adaptive Bins via Global Attention at High Resolution. KAUST Research Repository. https://doi.org/10.25781/KAUST-AC1MY
dc.identifier.doi10.25781/KAUST-AC1MY
dc.identifier.urihttp://hdl.handle.net/10754/668894
dc.description.abstractWe address the problem of estimating a high quality dense depth map from a single RGB input image. We start out with a baseline encoder-decoder convolutional neural network architecture and pose the question of how the global processing of information can help improve overall depth estimation. To this end, we propose a transformer-based architecture block that divides the depth range into bins whose center value is estimated adaptively per image. The final depth values are estimated as linear combinations of the bin centers. We call our new building block AdaBins. Our results show a decisive improvement over the state-of-the-art on several popular depth datasets across all metrics. We also validate the effectiveness of the proposed block with an ablation study.
dc.language.isoen
dc.subjectMonocular Depth Estimation
dc.subject3D reconstruction
dc.subjectTransformers
dc.subject3D scene understanding
dc.subjectadaptive binning
dc.subjectConvolutional Neural Networks
dc.titleDepth Estimation Using Adaptive Bins via Global Attention at High Resolution
dc.typeThesis
dc.contributor.departmentComputer, Electrical and Mathematical Science and Engineering (CEMSE) Division
thesis.degree.grantorKing Abdullah University of Science and Technology
dc.contributor.committeememberHadwiger, Markus
dc.contributor.committeememberGhanem, Bernard
thesis.degree.disciplineComputer Science
thesis.degree.nameMaster of Science
refterms.dateFOA2021-04-22T07:58:32Z
kaust.request.doiyes


Files in this item

Thumbnail
Name:
ShariqBhatThesis.pdf
Size:
27.30Mb
Format:
PDF

This item appears in the following Collection(s)

Show simple item record