Scholarlink

<p>The pure-vision multi-view 3D detection task is crucial for autonomous driving, necessitating precise environmental perception. Current methods fall short in effectively leveraging spatiotemporal information, often neglecting the differentiation and utilization of temporal information at various intervals and underutilizing multi-scale spatial information. To address these challenges, we introduce CrossBEV, a novel multi-view 3D perception framework that enhances spatial perception through cross-processing of historical frame features and reuse of multi-scale spatial features. CrossBEV comprises two important modules: Dual-Temporal Predictive Module (DTPM) and Cross-Frame Feature Reactivation (CFFR) Module. The DTPM module classifies historical frames for temporal processing, generating unified historical query features as pseudo-current frame features to constrain network training for object detection. The CFFR module integrates multi-scale image features across frames, boosting dynamic scene comprehension. Evaluated on the nuScenes dataset, CrossBEV achieved 65.1% NDS and 57.8% mAP in the test set, surpassing most 3D detectors. This approach provides a novel solution for 3D perception in autonomous driving by offering a new feature-enhancement architecture for multi-scale spatiotemporal feature utilization. Codes are available at <a href="https://github.com/AuRa-99/CrossBEV.">https://github.com/AuRa-99/CrossBEV.</a></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

CrossBEV: enhancing multi-view 3D object detection via spatiotemporal feature cross-enhancement

  • 期刊论文

  • 发表日期：

    2026年3月26日

查看全文
  • Jinlei Yu,
  • Junyin Wang,
  • Wenqian Zhu,
  • Huikai Liu,
  • Weidong Yang

摘要

The pure-vision multi-view 3D detection task is crucial for autonomous driving, necessitating precise environmental perception. Current methods fall short in effectively leveraging spatiotemporal information, often neglecting the differentiation and utilization of temporal information at various intervals and underutilizing multi-scale spatial information. To address these challenges, we introduce CrossBEV, a novel multi-view 3D perception framework that enhances spatial perception through cross-processing of historical frame features and reuse of multi-scale spatial features. CrossBEV comprises two important modules: Dual-Temporal Predictive Module (DTPM) and Cross-Frame Feature Reactivation (CFFR) Module. The DTPM module classifies historical frames for temporal processing, generating unified historical query features as pseudo-current frame features to constrain network training for object detection. The CFFR module integrates multi-scale image features across frames, boosting dynamic scene comprehension. Evaluated on the nuScenes dataset, CrossBEV achieved 65.1% NDS and 57.8% mAP in the test set, surpassing most 3D detectors. This approach provides a novel solution for 3D perception in autonomous driving by offering a new feature-enhancement architecture for multi-scale spatiotemporal feature utilization. Codes are available at https://github.com/AuRa-99/CrossBEV.