MonoDETRNext: Next-Generation Accurate and Efficient Monocular 3D Object Detector

TL;DR

MonoDETRNext introduces a new, more accurate and efficient monocular 3D object detection model, improving performance and speed over previous methods through enhanced architecture and depth estimation techniques.

Contribution

The paper presents MonoDETRNext, a novel DETR-based monocular 3D detector with two variants optimized for accuracy and speed, setting new benchmarks in the field.

Findings

MonoDETRNext-A improves AP3D by 3.52% on KITTI.

MonoDETRNext-E increases efficiency and AP3D by 2.35%.

The model outperforms existing solutions in accuracy and efficiency.

Abstract

Monocular 3D object detection has vast application potential across various fields. DETR-type models have shown remarkable performance in different areas, but there is still considerable room for improvement in monocular 3D detection, especially with the existing DETR-based method, MonoDETR. After addressing the query initialization issues in MonoDETR, we explored several performance enhancement strategies, such as incorporating a more efficient encoder and utilizing a more powerful depth estimator. Ultimately, we proposed MonoDETRNext, a model that comes in two variants based on the choice of depth estimator: MonoDETRNext-E, which prioritizes speed, and MonoDETRNext-A, which focuses on accuracy. We posit that MonoDETRNext establishes a new benchmark in monocular 3D object detection and opens avenues for future research. We conducted an exhaustive evaluation demonstrating the model's superior performance against existing solutions. Notably, MonoDETRNext-A demonstrated a 3.52$\%$ improvement in the $AP_{3D}$ metric on the KITTI test benchmark over MonoDETR, while MonoDETRNext-E showed a 2.35$\%$ increase. Additionally, the computational efficiency of MonoDETRNext-E slightly exceeds that of its predecessor.