New Adaptive Cache Method Boosts Rare Human-Object Interaction Detection

Researchers have developed an Adaptive Diversity Cache (ADC) framework to improve human-object interaction (HOI) detection, particularly for rare interactions. This training-free method adaptively allocates memory capacity during testing, prioritizing underrepresented interactions through frequency-dependent allocation and a confidence-diversity joint selection process. Experiments on HICO-DET and V-COCO datasets show significant improvements in mean average precision (mAP) for rare interactions, with gains ranging from 4.02 to 8.57 percentage points. AI

IMPACT Improves accuracy in detecting rare human-object interactions, potentially enhancing AI systems that rely on detailed scene understanding.

RANK_REASON This is a research paper detailing a new method for computer vision tasks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Adaptive Diversity Cache
• Confidence-Diversity Joint Cache Selection
• Frequency-Aware Cache Adaptation
• HICO-DET
• Yuqiu Jiang

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →