Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313183732> ?p ?o ?g. }
Showing items 1 to 94 of
94
with 100 items per page.
- W4313183732 endingPage "316" @default.
- W4313183732 startingPage "300" @default.
- W4313183732 abstract "High-resolution images are widely adopted for high-performance object detection in videos. However, processing high-resolution inputs comes with high computation costs, and naive down-sampling of the input to reduce the computation costs quickly degrades the detection performance. In this paper, we propose SALISA, a novel non-uniform SALiency-based Input SAmpling technique for video object detection that allows for heavy down-sampling of unimportant background regions while preserving the fine-grained details of a high-resolution image. The resulting image is spatially smaller, leading to reduced computational costs while enabling a performance comparable to a high-resolution input. To achieve this, we propose a differentiable resampling module based on a thin plate spline spatial transformer network (TPS-STN). This module is regularized by a novel loss to provide an explicit supervision signal to learn to “magnify” salient regions. We report state-of-the-art results in the low compute regime on the ImageNet-VID and UA-DETRAC video object detection datasets. We demonstrate that on both datasets, the mAP of an EfficientDet-D1 (EfficientDet-D2) gets on par with EfficientDet-D2 (EfficientDet-D3) at a much lower computational cost. We also show that SALISA significantly improves the detection of small objects. In particular, SALISA with an EfficientDet-D1 detector improves the detection of small objects by 77%, and remarkably also outperforms EfficientDet-D3 baseline." @default.
- W4313183732 created "2023-01-06" @default.
- W4313183732 creator A5004079118 @default.
- W4313183732 creator A5038550641 @default.
- W4313183732 creator A5045620960 @default.
- W4313183732 creator A5055531734 @default.
- W4313183732 date "2022-01-01" @default.
- W4313183732 modified "2023-10-17" @default.
- W4313183732 title "SALISA: Saliency-Based Input Sampling for Efficient Video Object Detection" @default.
- W4313183732 cites W1861492603 @default.
- W4313183732 cites W2018377917 @default.
- W4313183732 cites W2047957240 @default.
- W4313183732 cites W2117539524 @default.
- W4313183732 cites W2130859329 @default.
- W4313183732 cites W2549139847 @default.
- W4313183732 cites W2552900565 @default.
- W4313183732 cites W2604195031 @default.
- W4313183732 cites W2765877758 @default.
- W4313183732 cites W2889469641 @default.
- W4313183732 cites W2901155229 @default.
- W4313183732 cites W2963407932 @default.
- W4313183732 cites W2963653352 @default.
- W4313183732 cites W2963897760 @default.
- W4313183732 cites W2963948108 @default.
- W4313183732 cites W2979634944 @default.
- W4313183732 cites W2981669902 @default.
- W4313183732 cites W2981858655 @default.
- W4313183732 cites W2990152177 @default.
- W4313183732 cites W3003490096 @default.
- W4313183732 cites W3033613646 @default.
- W4313183732 cites W3034971973 @default.
- W4313183732 cites W3096609285 @default.
- W4313183732 cites W3106250896 @default.
- W4313183732 cites W3106753828 @default.
- W4313183732 cites W3109550678 @default.
- W4313183732 cites W3173421220 @default.
- W4313183732 cites W3198001539 @default.
- W4313183732 cites W4379967825 @default.
- W4313183732 cites W764651262 @default.
- W4313183732 doi "https://doi.org/10.1007/978-3-031-20080-9_18" @default.
- W4313183732 hasPublicationYear "2022" @default.
- W4313183732 type Work @default.
- W4313183732 citedByCount "0" @default.
- W4313183732 crossrefType "book-chapter" @default.
- W4313183732 hasAuthorship W4313183732A5004079118 @default.
- W4313183732 hasAuthorship W4313183732A5038550641 @default.
- W4313183732 hasAuthorship W4313183732A5045620960 @default.
- W4313183732 hasAuthorship W4313183732A5055531734 @default.
- W4313183732 hasBestOaLocation W43131837322 @default.
- W4313183732 hasConcept C106131492 @default.
- W4313183732 hasConcept C11413529 @default.
- W4313183732 hasConcept C140779682 @default.
- W4313183732 hasConcept C150921843 @default.
- W4313183732 hasConcept C153180895 @default.
- W4313183732 hasConcept C154945302 @default.
- W4313183732 hasConcept C205372480 @default.
- W4313183732 hasConcept C2776151529 @default.
- W4313183732 hasConcept C31972630 @default.
- W4313183732 hasConcept C41008148 @default.
- W4313183732 hasConcept C45374587 @default.
- W4313183732 hasConcept C76155785 @default.
- W4313183732 hasConcept C94915269 @default.
- W4313183732 hasConceptScore W4313183732C106131492 @default.
- W4313183732 hasConceptScore W4313183732C11413529 @default.
- W4313183732 hasConceptScore W4313183732C140779682 @default.
- W4313183732 hasConceptScore W4313183732C150921843 @default.
- W4313183732 hasConceptScore W4313183732C153180895 @default.
- W4313183732 hasConceptScore W4313183732C154945302 @default.
- W4313183732 hasConceptScore W4313183732C205372480 @default.
- W4313183732 hasConceptScore W4313183732C2776151529 @default.
- W4313183732 hasConceptScore W4313183732C31972630 @default.
- W4313183732 hasConceptScore W4313183732C41008148 @default.
- W4313183732 hasConceptScore W4313183732C45374587 @default.
- W4313183732 hasConceptScore W4313183732C76155785 @default.
- W4313183732 hasConceptScore W4313183732C94915269 @default.
- W4313183732 hasLocation W43131837321 @default.
- W4313183732 hasLocation W43131837322 @default.
- W4313183732 hasOpenAccess W4313183732 @default.
- W4313183732 hasPrimaryLocation W43131837321 @default.
- W4313183732 hasRelatedWork W1497101000 @default.
- W4313183732 hasRelatedWork W2095705906 @default.
- W4313183732 hasRelatedWork W2754428891 @default.
- W4313183732 hasRelatedWork W2770726253 @default.
- W4313183732 hasRelatedWork W2782964878 @default.
- W4313183732 hasRelatedWork W2922421953 @default.
- W4313183732 hasRelatedWork W2980691083 @default.
- W4313183732 hasRelatedWork W3002270006 @default.
- W4313183732 hasRelatedWork W3046346900 @default.
- W4313183732 hasRelatedWork W4281560450 @default.
- W4313183732 isParatext "false" @default.
- W4313183732 isRetracted "false" @default.
- W4313183732 workType "book-chapter" @default.