Papers/2609.35955
🧪 Test?View on arXiv

HEIR: Learning Human-Entity Interactions with Functional Roles

Kratos Wen, Author 2, Author 3, Author 4, Author 5

multimodalevent compositioninteraction prediction
2609.35955
Builder Relevance
80%
1h ago

Abstract

The paper presents HEIR, a benchmark for understanding human-entity interactions by recovering participant-role sets in various contexts.

Reality Card

Core Claim

HEIR introduces a comprehensive benchmark for evaluating human-entity interactions, significantly improving the understanding of event composition in shared environments.

Method / Result

CoRISP outperforms 16 baselines by 2.87 and 3.82 Set mAP points on repeated-role events and shared-participant images, respectively.

Limitations

The reliance on specific datasets may limit generalizability and reproducibility across diverse scenarios.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.
← Back to all papers