FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3178826664> ?p ?o ?g. }

• W3178826664 abstract "Vanilla models for object detection and instance segmentation suffer from the heavy bias toward detecting frequent objects in the long-tailed setting. Existing methods address this issue mostly during training, e.g., by re-sampling or re-weighting. In this paper, we investigate a largely overlooked approach -- post-processing calibration of confidence scores. We propose NorCal, Normalized Calibration for long-tailed object detection and instance segmentation, a simple and straightforward recipe that reweighs the predicted scores of each class by its training sample size. We show that separately handling the background class and normalizing the scores over classes for each proposal are keys to achieving superior performance. On the LVIS dataset, NorCal can effectively improve nearly all the baseline models not only on rare classes but also on common and frequent classes. Finally, we conduct extensive analysis and ablation studies to offer insights into various modeling choices and mechanisms of our approach. Our code is publicly available at https://github.com/tydpan/NorCal/." @default.
• W3178826664 created "2021-07-19" @default.
• W3178826664 creator A5011391278 @default.
• W3178826664 creator A5017319429 @default.
• W3178826664 creator A5018066903 @default.
• W3178826664 creator A5019483722 @default.
• W3178826664 creator A5059448769 @default.
• W3178826664 creator A5065708799 @default.
• W3178826664 creator A5075518871 @default.
• W3178826664 creator A5075596275 @default.
• W3178826664 date "2021-07-05" @default.
• W3178826664 modified "2023-09-23" @default.
• W3178826664 title "On Model Calibration for Long-Tailed Object Detection and Instance Segmentation" @default.
• W3178826664 cites W1565402342 @default.
• W3178826664 cites W1598033630 @default.
• W3178826664 cites W1618905105 @default.
• W3178826664 cites W1861492603 @default.
• W3178826664 cites W2012942264 @default.
• W3178826664 cites W2034603029 @default.
• W3178826664 cites W2086791339 @default.
• W3178826664 cites W2102605133 @default.
• W3178826664 cites W2108598243 @default.
• W3178826664 cites W2115763357 @default.
• W3178826664 cites W2118978333 @default.
• W3178826664 cites W2168356304 @default.
• W3178826664 cites W2194775991 @default.
• W3178826664 cites W2250384498 @default.
• W3178826664 cites W2254249950 @default.
• W3178826664 cites W2339172597 @default.
• W3178826664 cites W2549139847 @default.
• W3178826664 cites W2565639579 @default.
• W3178826664 cites W2570343428 @default.
• W3178826664 cites W2725875209 @default.
• W3178826664 cites W2767106145 @default.
• W3178826664 cites W2783231089 @default.
• W3178826664 cites W2797977484 @default.
• W3178826664 cites W2950141105 @default.
• W3178826664 cites W2962933664 @default.
• W3178826664 cites W2963037989 @default.
• W3178826664 cites W2963150697 @default.
• W3178826664 cites W2963351448 @default.
• W3178826664 cites W2963691377 @default.
• W3178826664 cites W2964212410 @default.
• W3178826664 cites W2964236837 @default.
• W3178826664 cites W2971118045 @default.
• W3178826664 cites W2972006294 @default.
• W3178826664 cites W2982540584 @default.
• W3178826664 cites W2983943451 @default.
• W3178826664 cites W2988916019 @default.
• W3178826664 cites W2995197345 @default.
• W3178826664 cites W2997580736 @default.
• W3178826664 cites W3011722050 @default.
• W3178826664 cites W3018757597 @default.
• W3178826664 cites W3021199811 @default.
• W3178826664 cites W3034542661 @default.
• W3178826664 cites W3034858314 @default.
• W3178826664 cites W3034933032 @default.
• W3178826664 cites W3035342001 @default.
• W3178826664 cites W3035552357 @default.
• W3178826664 cites W3035709993 @default.
• W3178826664 cites W3045241369 @default.
• W3178826664 cites W3081607473 @default.
• W3178826664 cites W3097096317 @default.
• W3178826664 cites W3101812930 @default.
• W3178826664 cites W3104182862 @default.
• W3178826664 cites W3105054740 @default.
• W3178826664 cites W3106060611 @default.
• W3178826664 cites W3106250896 @default.
• W3178826664 cites W3122855191 @default.
• W3178826664 cites W3166228617 @default.
• W3178826664 cites W3166596953 @default.
• W3178826664 cites W3170019900 @default.
• W3178826664 cites W3171998492 @default.
• W3178826664 cites W3172122468 @default.
• W3178826664 cites W3172971995 @default.
• W3178826664 cites W3173611024 @default.
• W3178826664 cites W3176659256 @default.
• W3178826664 cites W3182635745 @default.
• W3178826664 cites W3203618576 @default.
• W3178826664 cites W3203770998 @default.
• W3178826664 cites W3204574775 @default.
• W3178826664 cites W639708223 @default.
• W3178826664 doi "https://doi.org/10.48550/arxiv.2107.02170" @default.
• W3178826664 hasPublicationYear "2021" @default.
• W3178826664 type Work @default.
• W3178826664 sameAs 3178826664 @default.
• W3178826664 citedByCount "0" @default.
• W3178826664 crossrefType "posted-content" @default.
• W3178826664 hasAuthorship W3178826664A5011391278 @default.
• W3178826664 hasAuthorship W3178826664A5017319429 @default.
• W3178826664 hasAuthorship W3178826664A5018066903 @default.
• W3178826664 hasAuthorship W3178826664A5019483722 @default.
• W3178826664 hasAuthorship W3178826664A5059448769 @default.
• W3178826664 hasAuthorship W3178826664A5065708799 @default.
• W3178826664 hasAuthorship W3178826664A5075518871 @default.
• W3178826664 hasAuthorship W3178826664A5075596275 @default.
• W3178826664 hasBestOaLocation W31788266641 @default.
• W3178826664 hasConcept C105795698 @default.
• W3178826664 hasConcept C111368507 @default.
• W3178826664 hasConcept C119857082 @default.

• next