English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Unsupervised domain adaptive object detection aims to learn a robust detectorin the domain shift circumstance, where the training (source) domain islabel-rich with bounding box annotations, while the testing (target) domain islabel-agnostic and the feature distributions between training and testingdomains are dissimilar or even totally different. In this paper, we propose agradient detach based stacked complementary losses (SCL) method that usesdetection losses as the primary objective, and cuts in several auxiliary lossesin different network stages accompanying with gradient detach training to learnmore discriminative representations. We argue that the prior methods mainlyleverage more loss functions for training but ignore the interaction ofdifferent losses and also the compatible training strategy (gradient detachupdating in our work). Thus, our proposed method is a more syncretic adaptationlearning process. We conduct comprehensive experiments on seven datasets, theresults demonstrate that our method performs favorably better than thestate-of-the-art methods by a significant margin. For instance, from Cityscapesto FoggyCityscapes, we achieve 37.9% mAP, outperforming the previous artStrong-Weak by 3.6%.

SCL: Towards Accurate Domain Adaptive Object Detection via Gradient Detach Based Stacked Complementary Losses