TY - GEN
T1 - GLAMD
T2 - 17th European Conference on Computer Vision, ECCV 2022
AU - Jang, Younho
AU - Shin, Wheemyung
AU - Kim, Jinbeom
AU - Woo, Simon
AU - Bae, Sung Ho
N1 - Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2022
Y1 - 2022
N2 - Knowledge distillation (KD) is a well-known model compression strategy to improve models’ performance with fewer parameters. However, recent KD approaches for object detection have faced two limitations. First, they distill nearby foreground regions, ignoring potentially useful background information. Second, they only consider global contexts, thereby the student model can hardly learn local details from the teacher model. To overcome such challenging issues, we propose a novel knowledge distillation method, GLAMD, distilling both global and local knowledge from the teacher. We divide the feature maps into several patches and apply an attention mechanism for both the entire feature area and each patch to extract the global context as well as local details simultaneously. Our method outperforms the state-of-the-art methods with 40.8 AP on COCO2017 dataset, which is 3.4 AP higher than the student model (ResNet50 based Faster R-CNN) and 0.7 AP higher than the previous global attention-based distillation method.
AB - Knowledge distillation (KD) is a well-known model compression strategy to improve models’ performance with fewer parameters. However, recent KD approaches for object detection have faced two limitations. First, they distill nearby foreground regions, ignoring potentially useful background information. Second, they only consider global contexts, thereby the student model can hardly learn local details from the teacher model. To overcome such challenging issues, we propose a novel knowledge distillation method, GLAMD, distilling both global and local knowledge from the teacher. We divide the feature maps into several patches and apply an attention mechanism for both the entire feature area and each patch to extract the global context as well as local details simultaneously. Our method outperforms the state-of-the-art methods with 40.8 AP on COCO2017 dataset, which is 3.4 AP higher than the student model (ResNet50 based Faster R-CNN) and 0.7 AP higher than the previous global attention-based distillation method.
KW - Knowledge distillation
KW - Object detection
UR - https://www.scopus.com/pages/publications/85144594457
U2 - 10.1007/978-3-031-20080-9_27
DO - 10.1007/978-3-031-20080-9_27
M3 - Conference contribution
AN - SCOPUS:85144594457
SN - 9783031200793
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 460
EP - 476
BT - Computer Vision – ECCV 2022 - 17th European Conference, Proceedings
A2 - Avidan, Shai
A2 - Brostow, Gabriel
A2 - Cissé, Moustapha
A2 - Farinella, Giovanni Maria
A2 - Hassner, Tal
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 23 October 2022 through 27 October 2022
ER -