Matches in SemOpenAlex for { <https://semopenalex.org/work/W3091444184> ?p ?o ?g. }
Showing items 1 to 82 of
82
with 100 items per page.
- W3091444184 abstract "In this paper, we propose a novel loss rescaling method for single-stage convolutional object detection by predicted uncertainty scores. In object detection tasks, the difficulty of each object differs, and most of the background samples are easier to distinguish than others. Equally treating such samples causes data imbalance, resulting in underperforming accuracy and low efficiency in training. To resolve this issue, our method relates prediction difficulty to Bayesian uncertainty estimates and prediction correctness. Subsequently, we modify the architecture of a single-stage object detector to propagate Bayesian uncertainty and use the uncertainty estimates as anchor wise weights to re-scale the training loss. The proposed method is evaluated on both image classification and object detection tasks. The experiment results demonstrate that our method achieves better accuracy than existing loss rescaling methods, and verifies the effectiveness of Bayesian uncertainty inference in improving model training." @default.
- W3091444184 created "2020-10-08" @default.
- W3091444184 creator A5001154049 @default.
- W3091444184 creator A5024991309 @default.
- W3091444184 creator A5043806671 @default.
- W3091444184 creator A5075269741 @default.
- W3091444184 date "2020-10-01" @default.
- W3091444184 modified "2023-10-16" @default.
- W3091444184 title "Loss Rescaling by Uncertainty Inference For Single-Stage Object Detection" @default.
- W3091444184 cites W1677182931 @default.
- W3091444184 cites W2031489346 @default.
- W3091444184 cites W2103328396 @default.
- W3091444184 cites W2108598243 @default.
- W3091444184 cites W2120419212 @default.
- W3091444184 cites W2194775991 @default.
- W3091444184 cites W2597787948 @default.
- W3091444184 cites W2796347433 @default.
- W3091444184 cites W2884561390 @default.
- W3091444184 cites W2913592748 @default.
- W3091444184 cites W2951595529 @default.
- W3091444184 cites W2963516811 @default.
- W3091444184 cites W2964059111 @default.
- W3091444184 cites W2991403827 @default.
- W3091444184 cites W3106250896 @default.
- W3091444184 cites W3118608800 @default.
- W3091444184 doi "https://doi.org/10.1109/icip40778.2020.9190845" @default.
- W3091444184 hasPublicationYear "2020" @default.
- W3091444184 type Work @default.
- W3091444184 sameAs 3091444184 @default.
- W3091444184 citedByCount "0" @default.
- W3091444184 crossrefType "proceedings-article" @default.
- W3091444184 hasAuthorship W3091444184A5001154049 @default.
- W3091444184 hasAuthorship W3091444184A5024991309 @default.
- W3091444184 hasAuthorship W3091444184A5043806671 @default.
- W3091444184 hasAuthorship W3091444184A5075269741 @default.
- W3091444184 hasConcept C107673813 @default.
- W3091444184 hasConcept C11413529 @default.
- W3091444184 hasConcept C119857082 @default.
- W3091444184 hasConcept C121332964 @default.
- W3091444184 hasConcept C124101348 @default.
- W3091444184 hasConcept C153180895 @default.
- W3091444184 hasConcept C154945302 @default.
- W3091444184 hasConcept C160234255 @default.
- W3091444184 hasConcept C2776151529 @default.
- W3091444184 hasConcept C2776214188 @default.
- W3091444184 hasConcept C2778755073 @default.
- W3091444184 hasConcept C2781238097 @default.
- W3091444184 hasConcept C41008148 @default.
- W3091444184 hasConcept C55439883 @default.
- W3091444184 hasConcept C62520636 @default.
- W3091444184 hasConceptScore W3091444184C107673813 @default.
- W3091444184 hasConceptScore W3091444184C11413529 @default.
- W3091444184 hasConceptScore W3091444184C119857082 @default.
- W3091444184 hasConceptScore W3091444184C121332964 @default.
- W3091444184 hasConceptScore W3091444184C124101348 @default.
- W3091444184 hasConceptScore W3091444184C153180895 @default.
- W3091444184 hasConceptScore W3091444184C154945302 @default.
- W3091444184 hasConceptScore W3091444184C160234255 @default.
- W3091444184 hasConceptScore W3091444184C2776151529 @default.
- W3091444184 hasConceptScore W3091444184C2776214188 @default.
- W3091444184 hasConceptScore W3091444184C2778755073 @default.
- W3091444184 hasConceptScore W3091444184C2781238097 @default.
- W3091444184 hasConceptScore W3091444184C41008148 @default.
- W3091444184 hasConceptScore W3091444184C55439883 @default.
- W3091444184 hasConceptScore W3091444184C62520636 @default.
- W3091444184 hasLocation W30914441841 @default.
- W3091444184 hasOpenAccess W3091444184 @default.
- W3091444184 hasPrimaryLocation W30914441841 @default.
- W3091444184 hasRelatedWork W13366983 @default.
- W3091444184 hasRelatedWork W2164350 @default.
- W3091444184 hasRelatedWork W221938 @default.
- W3091444184 hasRelatedWork W2777878 @default.
- W3091444184 hasRelatedWork W695875 @default.
- W3091444184 hasRelatedWork W760780 @default.
- W3091444184 hasRelatedWork W8021486 @default.
- W3091444184 hasRelatedWork W9329549 @default.
- W3091444184 hasRelatedWork W6346411 @default.
- W3091444184 hasRelatedWork W722847 @default.
- W3091444184 isParatext "false" @default.
- W3091444184 isRetracted "false" @default.
- W3091444184 magId "3091444184" @default.
- W3091444184 workType "article" @default.