MonoVQD: Monocular 3D Object Detection with Variational Query Denoising and Self-Distillation
This work addresses a central problem in autonomous driving and robotics by improving monocular 3D object detection, though it appears incremental as it builds upon existing DETR-based methods with novel enhancements.
The paper tackles the challenge of precisely localizing 3D objects from a single image in monocular 3D detection by introducing MonoVQD, a framework that integrates variational query denoising and self-distillation into a DETR-based architecture, achieving superior performance on the KITTI benchmark and demonstrating broad applicability on datasets like nuScenes.
Precisely localizing 3D objects from a single image constitutes a central challenge in monocular 3D detection. While DETR-like architectures offer a powerful paradigm, their direct application in this domain encounters inherent limitations, preventing optimal performance. Our work addresses these challenges by introducing MonoVQD, a novel framework designed to fundamentally advance DETR-based monocular 3D detection. We propose three main contributions. First, we propose the Mask Separated Self-Attention mechanism that enables the integration of the denoising process into a DETR architecture. This improves the stability of Hungarian matching to achieve a consistent optimization objective. Second, we present the Variational Query Denoising technique to address the gradient vanishing problem of conventional denoising methods, which severely restricts the efficiency of the denoising process. This explicitly introduces stochastic properties to mitigate this fundamental limitation and unlock substantial performance gains. Finally, we introduce a sophisticated self-distillation strategy, leveraging insights from later decoder layers to synergistically improve query quality in earlier layers, thereby amplifying the iterative refinement process. Rigorous experimentation demonstrates that MonoVQD achieves superior performance on the challenging KITTI monocular benchmark. Highlighting its broad applicability, MonoVQD's core components seamlessly integrate into other architectures, delivering significant performance gains even in multi-view 3D detection scenarios on the nuScenes dataset and underscoring its robust generalization capabilities.