Matches in SemOpenAlex for { <https://semopenalex.org/work/W4320837918> ?p ?o ?g. }
Showing items 1 to 63 of
63
with 100 items per page.
- W4320837918 abstract "Knowledge distillation is an effective method for model compression. However, it is still a challenging topic to apply knowledge distillation to detection tasks. There are two key points resulting in poor distillation performance for detection tasks. One is the serious imbalance between foreground and background features, another one is that small object lacks enough feature representation. To solve the above issues, we propose a new distillation method named dual relation knowledge distillation (DRKD), including pixel-wise relation distillation and instance-wise relation distillation. The pixel-wise relation distillation embeds pixel-wise features in the graph space and applies graph convolution to capture the global pixel relation. By distilling the global pixel relation, the student detector can learn the relation between foreground and background features, and avoid the difficulty of distilling features directly for the feature imbalance issue. Besides, we find that instance-wise relation supplements valuable knowledge beyond independent features for small objects. Thus, the instance-wise relation distillation is designed, which calculates the similarity of different instances to obtain a relation matrix. More importantly, a relation filter module is designed to highlight valuable instance relations. The proposed dual relation knowledge distillation is general and can be easily applied for both one-stage and two-stage detectors. Our method achieves state-of-the-art performance, which improves Faster R-CNN based on ResNet50 from 38.4% to 41.6% mAP and improves RetinaNet based on ResNet50 from 37.4% to 40.3% mAP on COCO 2017." @default.
- W4320837918 created "2023-02-15" @default.
- W4320837918 creator A5015563980 @default.
- W4320837918 creator A5028517540 @default.
- W4320837918 creator A5069765960 @default.
- W4320837918 creator A5075717206 @default.
- W4320837918 date "2023-02-11" @default.
- W4320837918 modified "2023-09-27" @default.
- W4320837918 title "Dual Relation Knowledge Distillation for Object Detection" @default.
- W4320837918 doi "https://doi.org/10.48550/arxiv.2302.05637" @default.
- W4320837918 hasPublicationYear "2023" @default.
- W4320837918 type Work @default.
- W4320837918 citedByCount "0" @default.
- W4320837918 crossrefType "posted-content" @default.
- W4320837918 hasAuthorship W4320837918A5015563980 @default.
- W4320837918 hasAuthorship W4320837918A5028517540 @default.
- W4320837918 hasAuthorship W4320837918A5069765960 @default.
- W4320837918 hasAuthorship W4320837918A5075717206 @default.
- W4320837918 hasBestOaLocation W43208379181 @default.
- W4320837918 hasConcept C124101348 @default.
- W4320837918 hasConcept C132525143 @default.
- W4320837918 hasConcept C138885662 @default.
- W4320837918 hasConcept C153180895 @default.
- W4320837918 hasConcept C154945302 @default.
- W4320837918 hasConcept C160633673 @default.
- W4320837918 hasConcept C178790620 @default.
- W4320837918 hasConcept C185592680 @default.
- W4320837918 hasConcept C204030448 @default.
- W4320837918 hasConcept C25343380 @default.
- W4320837918 hasConcept C2776401178 @default.
- W4320837918 hasConcept C41008148 @default.
- W4320837918 hasConcept C41895202 @default.
- W4320837918 hasConcept C80444323 @default.
- W4320837918 hasConceptScore W4320837918C124101348 @default.
- W4320837918 hasConceptScore W4320837918C132525143 @default.
- W4320837918 hasConceptScore W4320837918C138885662 @default.
- W4320837918 hasConceptScore W4320837918C153180895 @default.
- W4320837918 hasConceptScore W4320837918C154945302 @default.
- W4320837918 hasConceptScore W4320837918C160633673 @default.
- W4320837918 hasConceptScore W4320837918C178790620 @default.
- W4320837918 hasConceptScore W4320837918C185592680 @default.
- W4320837918 hasConceptScore W4320837918C204030448 @default.
- W4320837918 hasConceptScore W4320837918C25343380 @default.
- W4320837918 hasConceptScore W4320837918C2776401178 @default.
- W4320837918 hasConceptScore W4320837918C41008148 @default.
- W4320837918 hasConceptScore W4320837918C41895202 @default.
- W4320837918 hasConceptScore W4320837918C80444323 @default.
- W4320837918 hasLocation W43208379181 @default.
- W4320837918 hasOpenAccess W4320837918 @default.
- W4320837918 hasPrimaryLocation W43208379181 @default.
- W4320837918 hasRelatedWork W1965781815 @default.
- W4320837918 hasRelatedWork W2047953819 @default.
- W4320837918 hasRelatedWork W2090093270 @default.
- W4320837918 hasRelatedWork W2136485282 @default.
- W4320837918 hasRelatedWork W2269705005 @default.
- W4320837918 hasRelatedWork W2546871836 @default.
- W4320837918 hasRelatedWork W2546942002 @default.
- W4320837918 hasRelatedWork W2547748020 @default.
- W4320837918 hasRelatedWork W3043252291 @default.
- W4320837918 hasRelatedWork W3160284275 @default.
- W4320837918 isParatext "false" @default.
- W4320837918 isRetracted "false" @default.
- W4320837918 workType "article" @default.