Published May 31, 2024
| Version v1
Preprint
Open
Descriptor Impact on Multimodal 3D Retrieval
Creators
-
1.
Centre for Research and Technology Hellas
-
2.
Reykjavík University
Description
With the evolution of 3D tools, there is now plenty of 3D data for digital applications. This includes 3D retrieval, which seeks to manage such data across varied representations like point clouds, meshes, and multi-view images. However, efficiently utilizing these representations for retrieval poses a challenge. This paper evaluates different representations of each modality in uni-modal retrieval and explores optimal combinations for multimodal retrieval. Results indicate MuseHash's superiority in MAP metric, while CMCL excels in recall. This study expands existing research by providing insights into optimal representations and combinations for 3D retrieval.
Files
cbmi2024_zenodo_version.pdf
Files
(623.8 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:a52c3136766389d9fee32660a10abdca
|
623.8 kB | Preview Download |
Additional details
Dates
- Accepted
-
2024-05-31