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

Object proposal generation serves as a standard pre-processing step inVision-Language (VL) tasks (image captioning, visual question answering, etc.).The performance of object proposals generated for VL tasks is currentlyevaluated across all available annotations, a protocol that we show ismisaligned - higher scores do not necessarily correspond to improvedperformance on downstream VL tasks. Our work serves as a study of thisphenomenon and explores the effectiveness of semantic grounding to mitigate itseffects. To this end, we propose evaluating object proposals against only asubset of available annotations, selected by thresholding an annotationimportance score. Importance of object annotations to VL tasks is quantified byextracting relevant semantic information from text describing the image. Weshow that our method is consistent and demonstrates greatly improved alignmentwith annotations selected by image captioning metrics and human annotation whencompared against existing techniques. Lastly, we compare current detectors usedin the Scene Graph Generation (SGG) benchmark as a use case, which serves as anexample of when traditional object proposal evaluation techniques aremisaligned.

Towards Addressing the Misalignment of Object Proposal Evaluation for Vision-Language Tasks via Semantic Grounding