基于能量分离的未知目标检测用于开放世界目标检测
Detecting Unknown Objects via Energy-based Separation for Open World Object Detection
用 ETF 构造正交子空间、并同时利用已知/未知两个空间的能量分数来分离未知目标，这一思路可直接迁移到开放词表检测与 OOD 目标发现中的未知判别头设计；EKD 损失对旧/新分类器施加分离约束，也可用于缓解增量检测中的新旧类知识干扰。
原文依据 EUS leverages ETF-based geometric properties to create orthogonal subspaces, enabling cleaner separation between known and unknown object representations.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论本文研究开放世界目标检测（OWOD）问题。这一具有挑战性的场景要求检测器在无监督条件下识别未知目标的同时，增量地学习对已知目标进行分类且不发生遗忘。此前的 OWOD 方法增强了未知目标发现过程，并采用记忆回放来缓解灾难性遗忘。然而，由于现有方法在检测未知目标时严重依赖检测器对已知类的预测，它们难以有效地学习并识别未知目标的表征。此外，记忆回放在缓解旧类遗忘的同时，往往会牺牲新学到的类的知识。为解决这些局限，我们提出 DEUS（通过能量分离检测未知目标），这是一个应对开放世界目标检测挑战的新框架。DEUS 由等角紧框架（ETF）子空间未知分离（EUS）与基于能量的已知区分（EKD）损失两部分组成。EUS 利用基于 ETF 的几何性质构造正交子空间，从而在已知与未知目标表征之间实现更清晰的分离。与此前仅考虑已知空间的基于能量的方法不同，EUS 同时利用两个空间的能量，以更好地捕捉未知目标的独特模式。此外，EKD 损失强制先前分类器与当前分类器之间保持分离，从而在记忆回放过程中最小化先前类与新学类之间的知识干扰。我们在 OWOD 基准上对 DEUS 进行了充分验证，结果表明其在未知目标检测上取得了显著的性能提升，同时保持了有竞争力的已知类性能。
英文原文摘要 1698 字符
In this work, we tackle the problem of Open World Object Detection (OWOD). This challenging scenario requires the detector to incrementally learn to classify known objects without forgetting while identifying unknown objects without supervision. Previous OWOD methods have enhanced the unknown discovery process and employed memory replay to mitigate catastrophic forgetting. However, since existing methods heavily rely on the detector's known class predictions for detecting unknown objects, they struggle to effectively learn and recognize unknown object representations. Moreover, while memory replay mitigates forgetting of old classes, it often sacrifices the knowledge of newly learned classes. To resolve these limitations, we propose DEUS (Detecting Unknowns via energy-based Separation), a novel framework that addresses the challenges of Open World Object Detection. DEUS consists of Equiangular Tight Frame (ETF)-Subspace Unknown Separation (EUS) and an Energy-based Known Distinction (EKD) loss. EUS leverages ETF-based geometric properties to create orthogonal subspaces, enabling cleaner separation between known and unknown object representations. Unlike prior energy-based approaches that consider only the known space, EUS utilizes energies from both spaces to better capture distinct patterns of unknown objects. Furthermore, EKD loss enforces the separation between previous and current classifiers, thus minimizing knowledge interference between previous and newly learned classes during memory replay. We thoroughly validate DEUS on OWOD benchmarks, demonstrating outstanding performance improvements in unknown detection while maintaining competitive known class performance.