[논문 리뷰] Center-based 3D Object Detection and Tracking

Tianwei Yin, Xingyi Zhou외 1명|arXiv (Cornell University)|2020. 06. 19.

Advanced Neural Network Applications참고 문헌 64인용 수 67

한 줄 요약

CenterPoint는 3D 물체를 축 정렬 상자가 아닌 중심점(점)으로 표현, 탐지 및 추적하여 LiDAR 데이터에서 간단하고 빠르며 최첨단의 3D 탐지 및 추적을 가능하게 한다. 두 단계의 정제가 성능을 더 향상시킨다.

ABSTRACT

Three-dimensional objects are commonly represented as 3D boxes in a point-cloud. This representation mimics the well-studied image-based 2D bounding-box detection but comes with additional challenges. Objects in a 3D world do not follow any particular orientation, and box-based detectors have difficulties enumerating all orientations or fitting an axis-aligned bounding box to rotated objects. In this paper, we instead propose to represent, detect, and track 3D objects as points. Our framework, CenterPoint, first detects centers of objects using a keypoint detector and regresses to other attributes, including 3D size, 3D orientation, and velocity. In a second stage, it refines these estimates using additional point features on the object. In CenterPoint, 3D object tracking simplifies to greedy closest-point matching. The resulting detection and tracking algorithm is simple, efficient, and effective. CenterPoint achieved state-of-the-art performance on the nuScenes benchmark for both 3D detection and tracking, with 65.5 NDS and 63.8 AMOTA for a single model. On the Waymo Open Dataset, CenterPoint outperforms all previous single model method by a large margin and ranks first among all Lidar-only submissions. The code and pretrained models are available at https://github.com/tianweiy/CenterPoint.

연구 동기 및 목표

• 3D 탐지에서 방향성 및 박스 피팅 문제를 극복하기 위해 중심 기반 표현을 제안한다.
• 센터포인트 검출기가 중심을 탐지하고 전체 3D 포즈, 크기 및 속도로 회귀하는 두 단계를 제안한다.
• 물체 표면의 포인트 특징을 활용한 가벼운 두 번째 단계 정제를 도입한다.
• 간단한 탐욕적 가장 가까운 점 매칭으로 엔드-투-엔드 3D 탐지 및 추적을 시연한다.
• 효율적인 추론으로 Waymo Open Dataset와 nuScenes에서 최첨단 성능을 보여준다.

제안 방법

• 일반적인 3D 백본( VoxelNet 또는 PointPillars )을 사용하여 LiDAR 데이터로부터 맵 뷰 특징 맵 M을 생성한다.
• 각 중심에서 클래스별 중심 히트맵, 물체 크기, 서브-보셀 로컬라이제이션, 요 각도, 속도를 밀집 헤드로 예측한다.
• CenterNet에서 영감을 받은 키포인트 손실과 모든 객체 속성에 대한 회귀 헤드를 사용하여 학습한다.
• 1단계 예측을 M에서 중심-페이스 포인트 특징을 샘플링하고 이를 MLP로 처리하여 신뢰도 점수와 박스 정제를 출력하는 두 번째 단계로 정제한다.
• 예측된 2D 속도를 사용하여 중심점을 투사하고 가장 가까운 점 매칭을 통해 현재 탐지와 이전 프레임을 탐욕적으로 매칭하여 추적한다.

실험 결과

연구 질문

• RQ1 Can a center-based representation simplify 3D object detection and tracking compared to anchor-based bounding boxes in LiDAR data?
• RQ2 Does a two-stage center-point refinement improve 3D localization, orientation, and velocity estimation with minimal overhead?
• RQ3 How well does center-based detection support real-time 3D tracking without heavy motion models?
• RQ4 What is the performance impact of center-based detection on large-scale datasets like Waymo and nuScenes?
• RQ5 How does CenterPoint compare to anchor-based methods in terms of robustness to rotation and object size variations?

주요 결과

• Center-based CenterPoint improves 3D detection by replacing axis-aligned boxes with center points, yielding substantial mAP/mAPH gains across backbones and datasets.
• Two-stage CenterPoint with additional center-face features provides further performance gains with modest overhead (2-stage refinement adds about 6–7 ms in some setups).
• CenterPoint achieves state-of-the-art results on Waymo (vehicle mAPH and pedestrian mAPH at Level 2) and nuScenes (NDS and mAP), outperforming prior methods.
• CenterPoint tracking via velocity-aware greedy closest-point matching significantly improves AMOTA over Kalman-filter baselines, with up to 8.8 AMOTA gains on nuScenes and substantial improvements on Waymo.
• CenterPoint runs near real-time (e.g., 11 FPS on Waymo, 16 FPS on nuScenes) and maintains strong performance across small and highly rotated objects.