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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

In Active Domain Adaptation (ADA), one uses Active Learning (AL) to select asubset of images from the target domain, which are then annotated and used forsupervised domain adaptation (DA). Given the large performance gap betweensupervised and unsupervised DA techniques, ADA allows for an excellenttrade-off between annotation cost and performance. Prior art makes use ofmeasures of uncertainty or disagreement of models to identify `regions' to beannotated by the human oracle. However, these regions frequently comprise ofpixels at object boundaries which are hard and tedious to annotate. Hence, evenif the fraction of image pixels annotated reduces, the overall annotation timeand the resulting cost still remain high. In this work, we propose an ADAstrategy, which given a frame, identifies a set of classes that are hardest forthe model to predict accurately, thereby recommending semantically meaningfulregions to be annotated in a selected frame. We show that these set of `hard'classes are context-dependent and typically vary across frames, and whenannotated help the model generalize better. We propose two ADA techniques: theAnchor-based and Augmentation-based approaches to select complementary anddiverse regions in the context of the current training set. Our approachachieves 66.6 mIoU on GTA to Cityscapes dataset with an annotation budget of4.7% in comparison to 64.9 mIoU by MADA using 5% of annotations. Our techniquecan also be used as a decorator for any existing frame-based AL technique,e.g., we report 1.5% performance improvement for CDAL on Cityscapes using ourapproach.

Reducing Annotation Effort by Identifying and Labeling Contextually Diverse Classes for Semantic Segmentation Under Domain Shift