DeepFusion: Lidar-Camera Deep Fusion for Multi-Modal 3D Object Detection

TL;DR

DeepFusion introduces novel techniques for aligning and fusing lidar and camera features at a deep level, significantly improving multi-modal 3D object detection accuracy and robustness in autonomous driving.

Contribution

The paper proposes InverseAug and LearnableAlign methods for effective lidar-camera feature fusion, leading to state-of-the-art detection performance.

Findings

Improves baseline models' pedestrian detection APH by up to 8.9%.

Achieves state-of-the-art results on Waymo Open Dataset.

Demonstrates robustness against input corruptions and out-of-distribution data.

Abstract

Lidars and cameras are critical sensors that provide complementary information for 3D detection in autonomous driving. While prevalent multi-modal methods simply decorate raw lidar point clouds with camera features and feed them directly to existing 3D detection models, our study shows that fusing camera features with deep lidar features instead of raw points, can lead to better performance. However, as those features are often augmented and aggregated, a key challenge in fusion is how to effectively align the transformed features from two modalities. In this paper, we propose two novel techniques: InverseAug that inverses geometric-related augmentations, e.g., rotation, to enable accurate geometric alignment between lidar points and image pixels, and LearnableAlign that leverages cross-attention to dynamically capture the correlations between image and lidar features during fusion. Based on InverseAug and LearnableAlign, we develop a family of generic multi-modal 3D detection models named DeepFusion, which is more accurate than previous methods. For example, DeepFusion improves PointPillars, CenterPoint, and 3D-MAN baselines on Pedestrian detection for 6.7, 8.9, and 6.2 LEVEL_2 APH, respectively. Notably, our models achieve state-of-the-art performance on Waymo Open Dataset, and show strong model robustness against input corruptions and out-of-distribution data. Code will be publicly available at https://github.com/tensorflow/lingvo/tree/master/lingvo/.