Joint prototype and coefficient prediction for 3D instance segmentation
Abstract 3D instance segmentation is crucial for applications demanding comprehensive 3D scene understanding. Here, a novel method is introduced that simultaneously learns coefficients and prototypes. Employing an overcomplete sampling strategy, the method produces an overcomplete set of instance pr...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2024-03-01
|
| Series: | Electronics Letters |
| Subjects: | |
| Online Access: | https://doi.org/10.1049/ell2.13137 |
| _version_ | 1797272626787254272 |
|---|---|
| author | Remco Royen Leon Denis Adrian Munteanu |
| author_facet | Remco Royen Leon Denis Adrian Munteanu |
| author_sort | Remco Royen |
| collection | DOAJ |
| description | Abstract 3D instance segmentation is crucial for applications demanding comprehensive 3D scene understanding. Here, a novel method is introduced that simultaneously learns coefficients and prototypes. Employing an overcomplete sampling strategy, the method produces an overcomplete set of instance predictions, from which the optimal ones are selected through a Non‐Maximum Suppression (NMS) algorithm during inference. The obtained prototypes are visualizable and interpretable. The method demonstrates superior performance on S3DIS‐blocks, consistently outperforming existing methods in mRec and mPrec. Moreover, it operates 32.9% faster than the state‐of‐the‐art. Notably, with only 0.8% of the total inference time, the method exhibits an over 20‐fold reduction in the variance of inference time compared to existing methods. These attributes render the method well‐suited for practical applications requiring both rapid inference and high reliability. |
| first_indexed | 2024-03-07T14:32:03Z |
| format | Article |
| id | doaj.art-7e06c0a3cc7b4054add6487fee79e7e8 |
| institution | Directory Open Access Journal |
| issn | 0013-5194 1350-911X |
| language | English |
| last_indexed | 2024-03-07T14:32:03Z |
| publishDate | 2024-03-01 |
| publisher | Wiley |
| record_format | Article |
| series | Electronics Letters |
| spelling | doaj.art-7e06c0a3cc7b4054add6487fee79e7e82024-03-06T04:11:14ZengWileyElectronics Letters0013-51941350-911X2024-03-01605n/an/a10.1049/ell2.13137Joint prototype and coefficient prediction for 3D instance segmentationRemco Royen0Leon Denis1Adrian Munteanu2Department of Electronics and Informatics Vrije Universiteit Brussel Brussels BelgiumDepartment of Electronics and Informatics Vrije Universiteit Brussel Brussels BelgiumDepartment of Electronics and Informatics Vrije Universiteit Brussel Brussels BelgiumAbstract 3D instance segmentation is crucial for applications demanding comprehensive 3D scene understanding. Here, a novel method is introduced that simultaneously learns coefficients and prototypes. Employing an overcomplete sampling strategy, the method produces an overcomplete set of instance predictions, from which the optimal ones are selected through a Non‐Maximum Suppression (NMS) algorithm during inference. The obtained prototypes are visualizable and interpretable. The method demonstrates superior performance on S3DIS‐blocks, consistently outperforming existing methods in mRec and mPrec. Moreover, it operates 32.9% faster than the state‐of‐the‐art. Notably, with only 0.8% of the total inference time, the method exhibits an over 20‐fold reduction in the variance of inference time compared to existing methods. These attributes render the method well‐suited for practical applications requiring both rapid inference and high reliability.https://doi.org/10.1049/ell2.13137computer visionimage processingimage segmentationlearning (artificial intelligence)neural nets |
| spellingShingle | Remco Royen Leon Denis Adrian Munteanu Joint prototype and coefficient prediction for 3D instance segmentation Electronics Letters computer vision image processing image segmentation learning (artificial intelligence) neural nets |
| title | Joint prototype and coefficient prediction for 3D instance segmentation |
| title_full | Joint prototype and coefficient prediction for 3D instance segmentation |
| title_fullStr | Joint prototype and coefficient prediction for 3D instance segmentation |
| title_full_unstemmed | Joint prototype and coefficient prediction for 3D instance segmentation |
| title_short | Joint prototype and coefficient prediction for 3D instance segmentation |
| title_sort | joint prototype and coefficient prediction for 3d instance segmentation |
| topic | computer vision image processing image segmentation learning (artificial intelligence) neural nets |
| url | https://doi.org/10.1049/ell2.13137 |
| work_keys_str_mv | AT remcoroyen jointprototypeandcoefficientpredictionfor3dinstancesegmentation AT leondenis jointprototypeandcoefficientpredictionfor3dinstancesegmentation AT adrianmunteanu jointprototypeandcoefficientpredictionfor3dinstancesegmentation |