運用消失點引導全景訊息預測

Abstract

本論文旨在解決自動駕駛領域中，僅使用單目相機進行三維語義場景補全（Semantic Scene Completion, SSC）時所面臨的關鍵挑戰，特別是對遠距離與微小物件感知準確度不足的問題。現有方法在處理因透視投影而在影像中變得微小、特徵模糊的遠方物體時，常因注意力分散而導致性能下降，進而對行車安全構成潛在威脅。為解決此問題，本研究提出一個名為「消失點聚合器」（VanishingPoint Aggregator, VPA）的創新架構。該方法的核心觀察在於：於駕駛場景影像中，消失點周圍自然聚集了來自遠距離場景的重要視覺資訊。VPA 引入一種新型的「消失點查詢」，專門用以強化此關鍵區域的特徵提取；並透過跨來源注意力融合機制，將富含遠場細節的 VPQ 與擷取全域物件語義的「標準實例查詢」進行整合，進而構建出更具完整性與辨識力的場景特徵表徵。 本研究於兩個具代表性的公開資料集——SemanticKITTI 與 SSCBench-KITTI-360 上進行系統性實驗與分析。實驗結果顯示，所提出的 VPA 模型在多項指標上皆達成目前最佳水準，尤其在遠距離區域與如行人、交通號誌等安全關鍵的微小物件類別上，顯著提升預測準確率。上述成果證實了本方法在提升單目 SSC 任務中遠場感知能力方面的有效性，對強化自動駕駛系統的環境感知穩定性與整體安全性具備實質貢獻。

Semantic Scene Completion (SSC) aims to jointly predict semantic categories and 3D occupancy of a scene from coarse inputs, which is crucial for providing reliable perception in autonomous driving. In this paper, we enhance existing SSC models by unveiling the vanishing point region, specifically addressing challenges posed by tiny objects and voxels distant from the monocular camera. At the core of our method, we propose the Vanishing Point Aggregator (VPA) to prioritize features in high-density central areas. The proposed VPA seamlessly integrates the Vanishing Point Query (VPQ) with the vanilla instance query via a cross-attention fusion mechanism to refine feature representation. To evaluate the effectiveness of our method, we conduct comprehensive experiments on two standard SSC benchmarks and demonstrate that our method achieves SOTA performance. Our ap- proach significantly improves the performance across various semantic classes, including a notable gain of 0.37 mIoU on SemanticKITTI and 0.5 mIoU on SSCBench-KITTI-360 for tiny objects. Ablation studies further validate the efficacy of our innovative query fusion strategy, showcasing its capability in long-range predictions for SSC tasks.