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

Determining which image regions to concentrate on is critical forHuman-Object Interaction (HOI) detection. Conventional HOI detectors focus oneither detected human and object pairs or pre-defined interaction locations,which limits learning of the effective features. In this paper, we reformulateHOI detection as an adaptive set prediction problem, with this novelformulation, we propose an Adaptive Set-based one-stage framework (AS-Net) withparallel instances and interaction branches. To attain this, we map a trainableinteraction query set to an interaction prediction set with a transformer. Eachquery adaptively aggregates the interaction-relevant features from globalcontexts through multi-head co-attention. Besides, the training process issupervised adaptively by matching each ground truth with the interactionprediction. Furthermore, we design an effective instance-aware attention moduleto introduce instructive features from the instance branch into the interactionbranch. Our method outperforms previous state-of-the-art methods without anyextra human pose and language features on three challenging HOI detectiondatasets. Especially, we achieve over $31\%$ relative improvement on alarge-scale HICO-DET dataset. Code is available athttps://github.com/yoyomimi/AS-Net.

Reformulating HOI Detection as Adaptive Set Prediction