FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4361996346> ?p ?o ?g. }

• W4361996346 endingPage "5998" @default.
• W4361996346 startingPage "5986" @default.
• W4361996346 abstract "Object detection is usually solved by deploying one single prediction head including classification and localization branches to obtain the final results. Recently proposed works utilize several prediction heads in a cascade learning manner to improve the detection performance. Despite achieving promising performance, existing cascade learning manner methods still meet with two inconsistency issues. Firstly, most of them refine the bounding boxes in different prediction heads only by depending on the localization accuracy ( <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>i.e</i> ., IoU), while ignoring the inconsistency between classification confidence and localization accuracy. Moreover, simply increasing the IoU threshold by experience to select positive samples makes the inconsistency issue even worse. Secondly, little consideration has been paid on the feature inconsistency between detection-specific features from different prediction heads and detection-generalized ones from backbone model. The extracted feature from backbone model contains the general representation for the whole images. While prediction heads need to be carefully designed to have specific ability which contains more discriminative expressions for the two sub-tasks classification and regression. The different contexture representations of the output features from these two parts lead to the feature inconsistency between backbone model and prediction head in cascade learning architecture. To solve these two inconsistency issues, this paper proposes a novel cascade consistency learning method for one-stage detector. Specifically, a feature adaptation module is firstly developed to calibrate features from different prediction heads and backbone model for solving the feature inconsistency. Then, we design an automatic positive sample threshold selection strategy for further solve the inconsistency between the classification and localization predictions. Moreover, the quality of bounding boxes in cascade learning manner are evaluated by taking both the classification confidence and localization accuracy into consideration. Experiments on MS COCO show that our proposed cascade consistency learning manner (dubbed C <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>2</sup> L) can achieve clear improvement over counterparts based on several different one-stage detectors, while performing favorably against state-of-the-arts." @default.
• W4361996346 created "2023-04-05" @default.
• W4361996346 creator A5005562185 @default.
• W4361996346 creator A5018318136 @default.
• W4361996346 creator A5053203094 @default.
• W4361996346 creator A5065669986 @default.
• W4361996346 creator A5067777366 @default.
• W4361996346 date "2023-10-01" @default.
• W4361996346 modified "2023-10-06" @default.
• W4361996346 title "Fully Cascade Consistency Learning for One-Stage Object Detection" @default.
• W4361996346 cites W2108598243 @default.
• W4361996346 cites W2194775991 @default.
• W4361996346 cites W2295107390 @default.
• W4361996346 cites W2549139847 @default.
• W4361996346 cites W2565639579 @default.
• W4361996346 cites W2570343428 @default.
• W4361996346 cites W2601564443 @default.
• W4361996346 cites W2886904239 @default.
• W4361996346 cites W2896991173 @default.
• W4361996346 cites W2962731685 @default.
• W4361996346 cites W2963037989 @default.
• W4361996346 cites W2963150697 @default.
• W4361996346 cites W2963299996 @default.
• W4361996346 cites W2963351448 @default.
• W4361996346 cites W2963497947 @default.
• W4361996346 cites W2963786238 @default.
• W4361996346 cites W2963849369 @default.
• W4361996346 cites W2964241181 @default.
• W4361996346 cites W2964342346 @default.
• W4361996346 cites W2964444661 @default.
• W4361996346 cites W2982770724 @default.
• W4361996346 cites W2988452521 @default.
• W4361996346 cites W2989604896 @default.
• W4361996346 cites W3011606420 @default.
• W4361996346 cites W3016916515 @default.
• W4361996346 cites W3035396860 @default.
• W4361996346 cites W3035473155 @default.
• W4361996346 cites W3035694605 @default.
• W4361996346 cites W3098090606 @default.
• W4361996346 cites W3102494826 @default.
• W4361996346 cites W3102701618 @default.
• W4361996346 cites W3106250896 @default.
• W4361996346 cites W3107473354 @default.
• W4361996346 cites W3107867277 @default.
• W4361996346 cites W3108849448 @default.
• W4361996346 cites W3109381875 @default.
• W4361996346 cites W3133630855 @default.
• W4361996346 cites W3152846589 @default.
• W4361996346 cites W3164543136 @default.
• W4361996346 cites W3167308647 @default.
• W4361996346 cites W3171162369 @default.
• W4361996346 cites W3171660447 @default.
• W4361996346 cites W3172087149 @default.
• W4361996346 cites W3176081225 @default.
• W4361996346 cites W3191338014 @default.
• W4361996346 cites W3196024568 @default.
• W4361996346 cites W3206423893 @default.
• W4361996346 cites W4214489586 @default.
• W4361996346 cites W4214507171 @default.
• W4361996346 cites W4221079664 @default.
• W4361996346 cites W4285345667 @default.
• W4361996346 doi "https://doi.org/10.1109/tcsvt.2023.3263557" @default.
• W4361996346 hasPublicationYear "2023" @default.
• W4361996346 type Work @default.
• W4361996346 citedByCount "0" @default.
• W4361996346 crossrefType "journal-article" @default.
• W4361996346 hasAuthorship W4361996346A5005562185 @default.
• W4361996346 hasAuthorship W4361996346A5018318136 @default.
• W4361996346 hasAuthorship W4361996346A5053203094 @default.
• W4361996346 hasAuthorship W4361996346A5065669986 @default.
• W4361996346 hasAuthorship W4361996346A5067777366 @default.
• W4361996346 hasConcept C115961682 @default.
• W4361996346 hasConcept C119857082 @default.
• W4361996346 hasConcept C138885662 @default.
• W4361996346 hasConcept C147037132 @default.
• W4361996346 hasConcept C153180895 @default.
• W4361996346 hasConcept C154945302 @default.
• W4361996346 hasConcept C17744445 @default.
• W4361996346 hasConcept C185592680 @default.
• W4361996346 hasConcept C199539241 @default.
• W4361996346 hasConcept C2776151529 @default.
• W4361996346 hasConcept C2776359362 @default.
• W4361996346 hasConcept C2776401178 @default.
• W4361996346 hasConcept C2776436953 @default.
• W4361996346 hasConcept C2781238097 @default.
• W4361996346 hasConcept C34146451 @default.
• W4361996346 hasConcept C41008148 @default.
• W4361996346 hasConcept C41895202 @default.
• W4361996346 hasConcept C43617362 @default.
• W4361996346 hasConcept C52622490 @default.
• W4361996346 hasConcept C59404180 @default.
• W4361996346 hasConcept C63584917 @default.
• W4361996346 hasConcept C94625758 @default.
• W4361996346 hasConcept C97931131 @default.
• W4361996346 hasConceptScore W4361996346C115961682 @default.
• W4361996346 hasConceptScore W4361996346C119857082 @default.
• W4361996346 hasConceptScore W4361996346C138885662 @default.
• W4361996346 hasConceptScore W4361996346C147037132 @default.

• next