Difficulty-Aware Label-Guided Denoising for Monocular 3D Object Detection
By: Soyul Lee, Seungmin Baek, Dongbo Min
Potential Business Impact:
Helps cars see better in 3D, even when objects are hidden.
Monocular 3D object detection is a cost-effective solution for applications like autonomous driving and robotics, but remains fundamentally ill-posed due to inherently ambiguous depth cues. Recent DETR-based methods attempt to mitigate this through global attention and auxiliary depth prediction, yet they still struggle with inaccurate depth estimates. Moreover, these methods often overlook instance-level detection difficulty, such as occlusion, distance, and truncation, leading to suboptimal detection performance. We propose MonoDLGD, a novel Difficulty-Aware Label-Guided Denoising framework that adaptively perturbs and reconstructs ground-truth labels based on detection uncertainty. Specifically, MonoDLGD applies stronger perturbations to easier instances and weaker ones into harder cases, and then reconstructs them to effectively provide explicit geometric supervision. By jointly optimizing label reconstruction and 3D object detection, MonoDLGD encourages geometry-aware representation learning and improves robustness to varying levels of object complexity. Extensive experiments on the KITTI benchmark demonstrate that MonoDLGD achieves state-of-the-art performance across all difficulty levels.
Similar Papers
Generalizing Monocular 3D Object Detection
CV and Pattern Recognition
Helps cars see in 3D from one picture.
DQ3D: Depth-guided Query for Transformer-Based 3D Object Detection in Traffic Scenarios
CV and Pattern Recognition
Helps cars see hidden objects better.
IDEAL-M3D: Instance Diversity-Enriched Active Learning for Monocular 3D Detection
CV and Pattern Recognition
Teaches AI to see 3D objects with fewer pictures.