MapTCL：通过双向对齐实现时间一致性学习以构建矢量化高清地图
原标题：MapTCL: Temporal Consistency Learning via Bidirectional Alignment for Vectorized HD Map Construction
AI 摘要
MapTCL 是一种用于矢量化高清地图构建的辅助训练策略，通过双向对齐实现时间一致性学习。它引入双向矢量一致性学习和栅格地图一致性学习两种损失，以惩罚连续在线高清地图之间的几何噪声和时间抖动。在 nuScenes 和 Argoverse 2 基准上，MapTCL 作为即插即用模块，分别提升了基线模型 +3.7 mAP 和 +3.1 mAP，且不增加推理开销。
正文节选
Computer Science > Computer Vision and Pattern Recognition Title:MapTCL: Temporal Consistency Learning via Bidirectional Alignment for Vectorized HD Map Construction View PDF HTML (experimental) Abstract:Constructing reliable online HD maps remains challenging in dynamic urban environments due to moving objects and occlusions. While recent works employ feature-level temporal fusion to address this, they rely solely on per-frame ground truth supervision. Consequently, they lack an exp