Every Pixel Matters: Center-Aware Feature Alignment for Domain Adaptive Object Detector

Cheng Chun Hsu*, Yi Hsuan Tsai, Yen Yu Lin, Ming Hsuan Yang

*此作品的通信作者

研究成果: Conference contribution同行評審

105 引文 斯高帕斯(Scopus)

摘要

A domain adaptive object detector aims to adapt itself to unseen domains that may contain variations of object appearance, viewpoints or backgrounds. Most existing methods adopt feature alignment either on the image level or instance level. However, image-level alignment on global features may tangle foreground/background pixels at the same time, while instance-level alignment using proposals may suffer from the background noise. Different from existing solutions, we propose a domain adaptation framework that accounts for each pixel via predicting pixel-wise objectness and centerness. Specifically, the proposed method carries out center-aware alignment by paying more attention to foreground pixels, hence achieving better adaptation across domains. We demonstrate our method on numerous adaptation settings with extensive experimental results and show favorable performance against existing state-of-the-art algorithms. Source codes and models are available at https://github.com/chengchunhsu/EveryPixelMatters.

原文English
主出版物標題Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings
編輯Andrea Vedaldi, Horst Bischof, Thomas Brox, Jan-Michael Frahm
發行者Springer Science and Business Media Deutschland GmbH
頁面733-748
頁數16
ISBN(列印)9783030585440
DOIs
出版狀態Published - 2020
事件16th European Conference on Computer Vision, ECCV 2020 - Glasgow, United Kingdom
持續時間: 23 8月 202028 8月 2020

出版系列

名字Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
12354 LNCS
ISSN(列印)0302-9743
ISSN(電子)1611-3349

Conference

Conference16th European Conference on Computer Vision, ECCV 2020
國家/地區United Kingdom
城市Glasgow
期間23/08/2028/08/20

指紋

深入研究「Every Pixel Matters: Center-Aware Feature Alignment for Domain Adaptive Object Detector」主題。共同形成了獨特的指紋。

引用此