EAN-MapNet: Efficient Vectorized HD Map Construction with Anchor Neighborhoods
This work addresses a specific bottleneck in autonomous driving systems by improving the efficiency and accuracy of HD map construction, representing an incremental advancement over existing DETR-based methods.
The paper tackles the problem of inefficient high-definition map construction for autonomous driving by proposing EAN-MapNet, which uses anchor neighborhoods and grouped local self-attention to reduce computational complexity and memory usage, achieving a state-of-the-art 63.0 mAP on nuScenes, a 12.7 mAP improvement over MapTR.
High-definition (HD) map is crucial for autonomous driving systems. Most existing works design map elements detection heads based on the DETR decoder. However, the initial queries lack explicit incorporation of physical positional information, and vanilla self-attention entails high computational complexity. Therefore, we propose EAN-MapNet for Efficiently constructing HD map using Anchor Neighborhoods. Firstly, we design query units based on the anchor neighborhoods, allowing non-neighborhood central anchors to effectively assist in fitting the neighborhood central anchors to the target points representing map elements. Then, we propose grouped local self-attention (GL-SA) by leveraging the relative instance relationship among the queries. This facilitates direct feature interaction among queries of the same instances, while innovatively employing local queries as intermediaries for interaction among queries from different instances. Consequently, GL-SA significantly reduces the computational complexity of self-attention while ensuring ample feature interaction among queries. On the nuScenes dataset, EAN-MapNet achieves a state-of-the-art performance with 63.0 mAP after training for 24 epochs, surpassing MapTR by 12.7 mAP. Furthermore, it considerably reduces memory consumption by 8198M compared to MapTRv2.