PDR: Progressive Depth Regularization for Monocular 3D Object Detection
- 15 May 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 33 (12) , 7591-7603
Abstract
Accurately predicting object depth is a key challenge in monocular 3D detection task. The perspective projection principle used by most state-of-the-art approaches demands a complex balance between the ratio-form depth estimation and 2D-3D geometric regularizations, and thus can lead to sub-optimal solutions. In this paper, we propose a novel synergistic scheme that can achieve better trade-off among these competing objectives. Our main proposal is a progressive depth regularization (PDR) architecture that splits the overall training process into three sequential depth estimation steps to gradually remove the unwanted deviations induced by the over-regularization. Specifically, our model first learns the coarse depth with the conventional perspective projection and combines the coarse-to-fine generation to reduce the search space of 2D projection height prediction. We then deactivate individual supervision on 2D projection height prediction and introduces a new auxiliary 3D physical height prediction to relax the 2D and 3D regularizations, respectively. Consequently, our PDR leads to more precise depth estimation by mitigating the inherent ambiguities in the geometric priors of perspective projection through progressive regularization relaxation. Extensive experiments on both KITTI and Rope3D benchmark show that our PDR delivers strong performance gains as compared to the previous methods.Keywords
Funding Information
- Zhejiang Provincial Key Laboratory of Information Processing, Communication and Networking (IPCAN), Hangzhou, China
- Fundamental Research Funds for the Central Universities (226-2022-00195)
- China Scholarship Council (202206320286)
- National Key Research and Development Program of China (2020AAA0103901)
- National Research Foundation, Singapore, under its Competitive Research Program Award (NRF-CRP23-2019-0003)
This publication has 58 references indexed in Scilit:
- SECOND: Sparsely Embedded Convolutional DetectionSensors, 2018
- Estimating Depth From Monocular Images as Classification Using Deep Fully Convolutional Residual NetworksIEEE Transactions on Circuits and Systems for Video Technology, 2017
- Densely Connected Convolutional Networks2017
- Feature Pyramid Networks for Object Detection2017
- PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation2017
- Multiple Instance Detection Network with Online Instance Classifier Refinement2017
- 3D Bounding Box Estimation Using Deep Learning and Geometry2017
- A Unified Multi-scale Deep Convolutional Neural Network for Fast Object Detection2016
- Deep Residual Learning for Image Recognition2016
- Vision meets robotics: The KITTI datasetThe International Journal of Robotics Research, 2013