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

Background subtraction (BGS) is a fundamental video processing task which isa key component of many applications. Deep learning-based supervised algorithmsachieve very good perforamnce in BGS, however, most of these algorithms areoptimized for either a specific video or a group of videos, and theirperformance decreases dramatically when applied to unseen videos. Recently,several papers addressed this problem and proposed video-agnostic supervisedBGS algorithms. However, nearly all of the data augmentations used in thesealgorithms are limited to the spatial domain and do not account for temporalvariations that naturally occur in video data. In this work, we introducespatio-temporal data augmentations and apply them to one of the leadingvideo-agnostic BGS algorithms, BSUV-Net. We also introduce a newcross-validation training and evaluation strategy for the CDNet-2014 datasetthat makes it possible to fairly and easily compare the performance of variousvideo-agnostic supervised BGS algorithms. Our new model trained using theproposed data augmentations, named BSUV-Net 2.0, significantly outperformsstate-of-the-art algorithms evaluated on unseen videos of CDNet-2014. We alsoevaluate the cross-dataset generalization capacity of BSUV-Net 2.0 by trainingit solely on CDNet-2014 videos and evaluating its performance on LASIESTAdataset. Overall, BSUV-Net 2.0 provides a ~5% improvement in the F-score overstate-of-the-art methods on unseen videos of CDNet-2014 and LASIESTA datasets.Furthermore, we develop a real-time variant of our model, that we call FastBSUV-Net 2.0, whose performance is close to the state of the art.

BSUV-Net 2.0: Spatio-Temporal Data Augmentations for Video-Agnostic Supervised Background Subtraction