.png){kind=icon}
.png){kind=icon}
Return
Efficient human–object-interaction (EHOI) detection via interaction label coding and Conditional Decision
.png){kind=icon}
PRE
AI
DOI：10.1016/j.cviu.2025.104390.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• EHOI is composed of feedforward modules, providing interpretable decisions. • EHOI offers comparable mAP performance with signi cantly lower inference complexity. • EHOI is explainable. All of its modules can be interpreted as probability estimators. • We formulate the learning process as the aggregation of conditional probabilities. • We explain modules’ decoupling statistically with visual examples. • We elaborate intermediate results, which verifies the relationship between modules.
Keywords:
EHOI
interpretable decisions
mAP performance
inference complexity
conditional probabilities
Journal
IF:
3.5
Papers:
428
Citations:
7.3K
