講演抄録/キーワード |
講演名 |
2020-12-17 15:10
Hierarchical Contrastive Adaptation for Cross-Domain Object Detection ○Ziwei Deng・Quan Kong・Naoto Akira・Tomoaki Yoshinaga(Hitachi) PRMU2020-46 |
抄録 |
(和) |
(まだ登録されていません) |
(英) |
Object detection based on deep learning has been enormously developed in recent years. However, applying detectors trained on a label-rich domain to an unseen domain results in performance drop due to the domain-shift. To deal with this problem, we propose a novel unsupervised domain adaptation framework to adapt the detector from a labeled source domain to an unlabeled target domain. In our proposed method, we utilize the image translation to generate interpolated images of source and target domains to fill in the large domain gap and facilitate a paired adaptation. We propose a hierarchical contrastive adaptation method between the original and interpolated domains to encourage the detectors to learn domain-invariant but discriminative features. To tackle the noises brought by image translation, we further propose a foreground attention reweighting for instance-aware adaptation. Experiments are carried out on 3 different scenarios of cross-domain detection and we achieve the state-of-the-art results against other approaches, showing the effectiveness of our proposed method. |
キーワード |
(和) |
/ / / / / / / |
(英) |
Unsupervised domain adaptation / Transfer learning / Object detection / / / / / |
文献情報 |
信学技報, vol. 120, no. 300, PRMU2020-46, pp. 47-52, 2020年12月. |
資料番号 |
PRMU2020-46 |
発行日 |
2020-12-10 (PRMU) |
ISSN |
Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
PDFダウンロード |
PRMU2020-46 |
|