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

Visual Grounding (VG) aims at localizing target objects from an image basedon given expressions and has made significant progress with the development ofdetection and vision transformer. However, existing VG methods tend to generatefalse-alarm objects when presented with inaccurate or irrelevant descriptions,which commonly occur in practical applications. Moreover, existing methods failto capture fine-grained features, accurate localization, and sufficient contextcomprehension from the whole image and textual descriptions. To address bothissues, we propose an Iterative Robust Visual Grounding (IR-VG) framework withMasked Reference based Centerpoint Supervision (MRCS). The framework introducesiterative multi-level vision-language fusion (IMVF) for better alignment. Weuse MRCS to ahieve more accurate localization with point-wised featuresupervision. Then, to improve the robustness of VG, we also present amulti-stage false-alarm sensitive decoder (MFSD) to prevent the generation offalse-alarm objects when presented with inaccurate expressions. The proposedframework is evaluated on five regular VG datasets and two newly constructedrobust VG datasets. Extensive experiments demonstrate that IR-VG achieves newstate-of-the-art (SOTA) results, with improvements of 25\% and 10\% compared toexisting SOTA approaches on the two newly proposed robust VG datasets.Moreover, the proposed framework is also verified effective on five regular VGdatasets. Codes and models will be publicly athttps://github.com/cv516Buaa/IR-VG.

Iterative Robust Visual Grounding with Masked Reference based Centerpoint Supervision