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

We introduce dense relational captioning, a novel image captioning task whichaims to generate multiple captions with respect to relational informationbetween objects in a visual scene. Relational captioning provides explicitdescriptions for each relationship between object combinations. This frameworkis advantageous in both diversity and amount of information, leading to acomprehensive image understanding based on relationships, e.g., relationalproposal generation. For relational understanding between objects, thepart-of-speech (POS; i.e., subject-object-predicate categories) can be avaluable prior information to guide the causal sequence of words in a caption.We enforce our framework to learn not only to generate captions but also tounderstand the POS of each word. To this end, we propose the multi-tasktriple-stream network (MTTSNet) which consists of three recurrent unitsresponsible for each POS which is trained by jointly predicting the correctcaptions and POS for each word. In addition, we found that the performance ofMTTSNet can be improved by modulating the object embeddings with an explicitrelational module. We demonstrate that our proposed model can generate morediverse and richer captions, via extensive experimental analysis on large scaledatasets and several metrics. Then, we present applications of our framework toholistic image captioning, scene graph generation, and retrieval tasks.

Dense Relational Image Captioning via Multi-task Triple-Stream Networks