Object detection with 3D radar is essential for 360-degree automotive perception, but radar's long wavelengths produce sparse and irregular reflections that challenge traditional grid and sequence-based convolutional and transformer detectors. This paper introduces Graph Query Networks (GQN), an attention-based framework that models objects sensed by radar as graphs, to extract individualized relational and contextual features. GQN employs a novel concept of graph queries to dynamically attend over the bird's-eye view (BEV) space, constructing object-specific graphs processed by two novel modules: EdgeFocus for relational reasoning and DeepContext Pooling for contextual aggregation. On the NuScenes dataset, GQN improves relative mAP by up to +53%, including a +8.2% gain over the strongest prior radar method, while reducing peak graph construction overhead by 80% with moderate FLOPs cost.

翻译：三维雷达目标检测对于实现360度汽车感知至关重要，但雷达的长波长特性会产生稀疏且不规则的反射信号，这对传统的基于网格和序列的卷积及Transformer检测器构成了挑战。本文提出图查询网络（Graph Query Networks, GQN），一种基于注意力的框架，将雷达感知的目标建模为图结构，以提取个体化的关系与上下文特征。GQN采用新颖的图查询概念，动态关注鸟瞰图（Bird's-Eye View, BEV）空间，构建由两个创新模块处理的目标专用图：用于关系推理的EdgeFocus模块和用于上下文聚合的DeepContext Pooling模块。在NuScenes数据集上，GQN将相对平均精度（mAP）提升高达+53%，包括对现有最强雷达方法额外提升+8.2%，同时以适度的浮点运算（FLOPs）成本将峰值图构建开销降低80%。