핵심 개념
提案されたDisentangled Object-Centric TRansformer(DOCTR)は、複数のオブジェクトと複数のサブタスクを統一的に学習することを可能にします。
통계
Qualitative experimental results demonstrate that our method achieves state-of-the-art performance on the challenging ScanNet dataset.
인용구
"Our method enables direct sparse predictions, yielding fewer false positives compared to existing methods."
"Extensive experiments demonstrate our superior performance than previous SOTA methods."