OPEN: Object-wise Position Embedding for Multi-view 3D Object Detection

TL;DR

The paper introduces OPEN, a multi-view 3D object detection method that incorporates object-wise depth information via position embedding, significantly improving detection accuracy on the nuScenes benchmark.

Contribution

The paper proposes a novel object-wise position embedding technique that effectively injects object-level depth information into transformer-based 3D detectors.

Findings

Achieves state-of-the-art 64.4% NDS on nuScenes.

Improves 3D detection accuracy by leveraging object-wise depth.

Demonstrates the effectiveness of object-wise depth encoding.

Abstract

Accurate depth information is crucial for enhancing the performance of multi-view 3D object detection. Despite the success of some existing multi-view 3D detectors utilizing pixel-wise depth supervision, they overlook two significant phenomena: 1) the depth supervision obtained from LiDAR points is usually distributed on the surface of the object, which is not so friendly to existing DETR-based 3D detectors due to the lack of the depth of 3D object center; 2) for distant objects, fine-grained depth estimation of the whole object is more challenging. Therefore, we argue that the object-wise depth (or 3D center of the object) is essential for accurate detection. In this paper, we propose a new multi-view 3D object detector named OPEN, whose main idea is to effectively inject object-wise depth information into the network through our proposed object-wise position embedding. Specifically, we first employ an object-wise depth encoder, which takes the pixel-wise depth map as a prior, to accurately estimate the object-wise depth. Then, we utilize the proposed object-wise position embedding to encode the object-wise depth information into the transformer decoder, thereby producing 3D object-aware features for final detection. Extensive experiments verify the effectiveness of our proposed method. Furthermore, OPEN achieves a new state-of-the-art performance with 64.4% NDS and 56.7% mAP on the nuScenes test benchmark.