Teach-DETR: Better Training DETR With Teachers
IEEE Transactions on Pattern Analysis and Machine Intelligence · 2023-12 · 45 (12) · 15759-15771
DOI: 10.1109/tpami.2023.3319387 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023-12. Verfasst von Linjiang Huang und 6 weiteren Autor:innen. Die Arbeit wurde bisher 16-mal zitiert (Crossref) und nennt 44 Literaturangaben. Gefördert von: National Key R&D Program of China Project, Centre for Perceptual and Interactive Intelligence, Innovation and Technology Commission (ITC)'s InnoHK.
Abstract (Originalsprache)
In this paper, we present a novel training scheme, namely Teach-DETR, to better train DETR-based detectors from versatile types of teacher detectors. We show that the predicted boxes from teacher detectors are effective medium to transfer knowledge of teacher detectors, which could be either RCNN-based or DETR-based detectors, to train a more accurate and robust DETR model. This new training scheme can easily incorporate the predicted boxes from multiple teacher detectors, each of which provides parallel supervisions to the student DETR. Our strategy introduces no additional parameters and adds negligible computational cost to the original detector during training. During inference, Teach-DETR brings zero additional overhead and maintains the merit of requiring no non-maximum suppression. Extensive experiments show that our method leads to consistent improvement for various DETR-based detectors. Specifically, we improve the state-of-the-art detector DINO Zhang et al. 2022 with Swin-Large Liu et al. 2021 backbone, 4-scale feature pyramid and 36-epoch training schedule, from 57.8% to 58.9% in terms of mean average precision on COCO 2017 val set.
Förderung
- National Key R&D Program of China Project
- Centre for Perceptual and Interactive Intelligence
- Innovation and Technology Commission (ITC)'s InnoHK
- General Research Fund of Hong Kong RGC Project
- PI of CPII
- InnoHK
Zitieren
APA
Huang, L., Lu, K., Song, G., Wang, L., Liu, S., Liu, Y., & Li, H. (2023). Teach-DETR: Better Training DETR With Teachers. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(12), 15759-15771. https://doi.org/10.1109/tpami.2023.3319387
Vancouver
Huang L, Lu K, Song G, Wang L, Liu S, Liu Y, et al. Teach-DETR: Better Training DETR With Teachers. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2023;45(12):15759-15771. doi:10.1109/tpami.2023.3319387
BibTeX
@article{Huang2023,
author = {Huang, Linjiang and Lu, Kaixin and Song, Guanglu and Wang, Liang and Liu, Si and Liu, Yu and Li, Hongsheng},
title = {Teach-DETR: Better Training DETR With Teachers},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
year = {2023},
volume = {45},
number = {12},
pages = {15759--15771},
doi = {10.1109/tpami.2023.3319387}
}
EUVIMED