Matches in SemOpenAlex for { <https://semopenalex.org/work/W4288889781> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4288889781 abstract "Class imbalance is a fundamental problem in computer vision applications such as semantic segmentation. Specifically, uneven class distributions in a training dataset often result in unsatisfactory performance on under-represented classes. Many works have proposed to weight the standard cross entropy loss function with pre-computed weights based on class statistics, such as the number of samples and class margins. There are two major drawbacks to these methods: 1) constantly up-weighting minority classes can introduce excessive false positives in semantic segmentation; 2) a minority class is not necessarily a hard class. The consequence is low precision due to excessive false positives. In this regard, we propose a hard-class mining loss by reshaping the vanilla cross entropy loss such that it weights the loss for each class dynamically based on instantaneous recall performance. We show that the novel recall loss changes gradually between the standard cross entropy loss and the inverse frequency weighted loss. Recall loss also leads to improved mean accuracy while offering competitive mean Intersection over Union (IoU) performance. On Synthia dataset, recall loss achieves $9%$ relative improvement on mean accuracy with competitive mean IoU using DeepLab-ResNet18 compared to the cross entropy loss. Code available at https://github.com/PotatoTian/recall-semseg." @default.
- W4288889781 created "2022-07-31" @default.
- W4288889781 creator A5012973327 @default.
- W4288889781 creator A5024074850 @default.
- W4288889781 creator A5036342749 @default.
- W4288889781 creator A5085681583 @default.
- W4288889781 creator A5088892134 @default.
- W4288889781 date "2021-06-28" @default.
- W4288889781 modified "2023-09-24" @default.
- W4288889781 title "Striking the Right Balance: Recall Loss for Semantic Segmentation" @default.
- W4288889781 doi "https://doi.org/10.48550/arxiv.2106.14917" @default.
- W4288889781 hasPublicationYear "2021" @default.
- W4288889781 type Work @default.
- W4288889781 citedByCount "0" @default.
- W4288889781 crossrefType "posted-content" @default.
- W4288889781 hasAuthorship W4288889781A5012973327 @default.
- W4288889781 hasAuthorship W4288889781A5024074850 @default.
- W4288889781 hasAuthorship W4288889781A5036342749 @default.
- W4288889781 hasAuthorship W4288889781A5085681583 @default.
- W4288889781 hasAuthorship W4288889781A5088892134 @default.
- W4288889781 hasBestOaLocation W42888897811 @default.
- W4288889781 hasConcept C100660578 @default.
- W4288889781 hasConcept C105795698 @default.
- W4288889781 hasConcept C106301342 @default.
- W4288889781 hasConcept C121332964 @default.
- W4288889781 hasConcept C126838900 @default.
- W4288889781 hasConcept C153180895 @default.
- W4288889781 hasConcept C154945302 @default.
- W4288889781 hasConcept C15744967 @default.
- W4288889781 hasConcept C167981619 @default.
- W4288889781 hasConcept C180747234 @default.
- W4288889781 hasConcept C183115368 @default.
- W4288889781 hasConcept C2777212361 @default.
- W4288889781 hasConcept C33923547 @default.
- W4288889781 hasConcept C41008148 @default.
- W4288889781 hasConcept C62520636 @default.
- W4288889781 hasConcept C64869954 @default.
- W4288889781 hasConcept C71924100 @default.
- W4288889781 hasConcept C89600930 @default.
- W4288889781 hasConceptScore W4288889781C100660578 @default.
- W4288889781 hasConceptScore W4288889781C105795698 @default.
- W4288889781 hasConceptScore W4288889781C106301342 @default.
- W4288889781 hasConceptScore W4288889781C121332964 @default.
- W4288889781 hasConceptScore W4288889781C126838900 @default.
- W4288889781 hasConceptScore W4288889781C153180895 @default.
- W4288889781 hasConceptScore W4288889781C154945302 @default.
- W4288889781 hasConceptScore W4288889781C15744967 @default.
- W4288889781 hasConceptScore W4288889781C167981619 @default.
- W4288889781 hasConceptScore W4288889781C180747234 @default.
- W4288889781 hasConceptScore W4288889781C183115368 @default.
- W4288889781 hasConceptScore W4288889781C2777212361 @default.
- W4288889781 hasConceptScore W4288889781C33923547 @default.
- W4288889781 hasConceptScore W4288889781C41008148 @default.
- W4288889781 hasConceptScore W4288889781C62520636 @default.
- W4288889781 hasConceptScore W4288889781C64869954 @default.
- W4288889781 hasConceptScore W4288889781C71924100 @default.
- W4288889781 hasConceptScore W4288889781C89600930 @default.
- W4288889781 hasLocation W42888897811 @default.
- W4288889781 hasOpenAccess W4288889781 @default.
- W4288889781 hasPrimaryLocation W42888897811 @default.
- W4288889781 hasRelatedWork W10575013 @default.
- W4288889781 hasRelatedWork W10828093 @default.
- W4288889781 hasRelatedWork W1163424 @default.
- W4288889781 hasRelatedWork W13608894 @default.
- W4288889781 hasRelatedWork W13815759 @default.
- W4288889781 hasRelatedWork W2651071 @default.
- W4288889781 hasRelatedWork W4434041 @default.
- W4288889781 hasRelatedWork W4797066 @default.
- W4288889781 hasRelatedWork W8258973 @default.
- W4288889781 hasRelatedWork W9388322 @default.
- W4288889781 isParatext "false" @default.
- W4288889781 isRetracted "false" @default.
- W4288889781 workType "article" @default.