CoIn3D：重新审视配置不变的多相机3D目标检测
CoIn3D: Revisiting Configuration-Invariant Multi-Camera 3D Object Detection
针对多相机3D检测在相机内参、外参和阵列布局变化下的跨平台泛化问题，提出将焦距、地面深度、地面梯度与 Plücker 坐标四类空间先验注入特征调制，并用免训练的动态新视角合成做相机感知数据增强；这套「空间先验显式建模 + 配置扰动增强」思路可直接迁移到任何需要跨传感器配置部署的 BEV/透视图 3D 感知模型。
原文依据 the devil lies in spatial prior discrepancies across source and target configurations, including different intrinsics, extrinsics, and array layouts
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论随着机器人与自动驾驶车辆等多传感器物理智能体的部署日益普及，多相机3D目标检测（MC3D）受到越来越多的关注。然而，MC3D 模型仍难以泛化到具有全新多相机配置的未见平台上。现有解决方案只是简单地采用一个元相机（meta-camera）来获得统一表示，但缺乏全面的考量。本文重新审视这一问题，并指出关键在于源配置与目标配置之间的空间先验差异，包括不同的内参、外参以及阵列布局。为此，我们提出 CoIn3D，一个具备强迁移能力的通用 MC3D 框架，能够从源配置强迁移到未见的目标配置。CoIn3D 分别通过空间感知特征调制（SFM）和相机感知数据增强（CDA），将上述所有识别出的空间先验显式地融入特征嵌入与图像观测之中。SFM 通过整合焦距、地面深度、地面梯度与 Plücker 坐标这四种空间表示来丰富特征空间。CDA 则借助一种免训练的动态新视角图像合成方案，提升模型在各种配置下的观测多样性。大量实验表明，在以 BEVDepth、BEVFormer 和 PETR 为代表的三种主流 MC3D 范式下，CoIn3D 在 NuScenes、Waymo 和 Lyft 等标志性数据集上均取得了较强的跨配置性能。
英文原文摘要 1474 字符
Multi-camera 3D object detection (MC3D) has attracted increasing attention with the growing deployment of multi-sensor physical agents, such as robots and autonomous vehicles. However, MC3D models still struggle to generalize to unseen platforms with new multi-camera configurations. Current solutions simply employ a meta-camera for unified representation but lack comprehensive consideration. In this paper, we revisit this issue and identify that the devil lies in spatial prior discrepancies across source and target configurations, including different intrinsics, extrinsics, and array layouts. To address this, we propose CoIn3D, a generalizable MC3D framework that enables strong transferability from source configurations to unseen target ones. CoIn3D explicitly incorporates all identified spatial priors into both feature embedding and image observation through spatial-aware feature modulation (SFM) and camera-aware data augmentation (CDA), respectively. SFM enriches feature space by integrating four spatial representations, such as focal length, ground depth, ground gradient, and Plucker coordinate. CDA improves observation diversity under various configurations via a training-free dynamic novel-view image synthesis scheme. Extensive experiments demonstrate that CoIn3D achieves strong cross-configuration performance on landmark datasets such as NuScenes, Waymo, and Lyft, under three dominant MC3D paradigms represented by BEVDepth, BEVFormer, and PETR.