• Home
  • Search
  • Autolabeling 3D Objects With Differentiable Rendering of SDF Shape Priors
  • Open Access IconOpen Access
  • Cite Icon114
  • https://doi.org/10.1109/cvpr42600.2020.01224Copy DOI Icon

Autolabeling 3D Objects With Differentiable Rendering of SDF Shape Priors

  • Jun 1, 2020
  • Sergey Zakharov +3 more
Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

We present an automatic annotation pipeline to recover 9D cuboids and 3D shapes from pre-trained off-the-shelf 2D detectors and sparse LIDAR data. Our autolabeling method solves an ill-posed inverse problem by considering learned shape priors and optimizing geometric and physical parameters. To address this challenging problem, we apply a novel differentiable shape renderer to signed distance fields (SDF), leveraged together with normalized object coordinate spaces (NOCS). Initially trained on synthetic data to predict shape and coordinates, our method uses these predictions for projective and geometric alignment over real samples. Moreover, we also propose a curriculum learning strategy, iteratively retraining on samples of increasing difficulty in subsequent self-improving annotation rounds. Our experiments on the KITTI3D dataset show that we can recover a substantial amount of accurate cuboids, and that these autolabels can be used to train 3D vehicle detectors with state-of-the-art results.

Similar Papers
  • Dissertation

Investigation on Segmentation, Recognition and 3D Reconstruction of Objects Based on LiDAR Data Or MRI

  • May 01, 2015
  • Shijun Tang
  • PDF
  • Research Article
  • Citations6

Model-Based Retrieval of Forest Parameters From Sentinel-1 Coherence and Backscatter Time Series

  • Jan 01, 2023
  • IEEE Geoscience and Remote Sensing Letters
  • M Lavalle +5
  • PDF
  • Research Article
  • Citations8

NNNet: New Normal Guided Depth Completion From Sparse LiDAR Data and Single Color Image

  • Jan 01, 2022
  • IEEE Access
  • Jiade Liu +1
  • Book Chapter
  • Citations4

Disentangling Deep Network for Reconstructing 3D Object Shapes from Single 2D Images

  • Jan 01, 2021
  • Yang Yang +3
  • PDF
  • Research Article
  • Citations46

Deep Architecture With Cross Guidance Between Single Image and Sparse LiDAR Data for Depth Completion

  • Jan 01, 2020
  • IEEE Access
  • Sihaeng Lee +3
  • Research Article
  • Citations2

Point Cloud Wall Projection for Realistic Road Data Augmentation.

  • Dec 20, 2024
  • Sensors (Basel, Switzerland)
  • Kana Kim +4
  • Research Article
  • Citations4

Spacecraft depth completion from sparse LiDAR data under adverse illumination

  • Aug 01, 2025
  • Aerospace Science and Technology
  • Ao Xiang +1
  • Conference Article

DG-BEV: Depth-Guided BEV 3D Object Detection with Sparse LiDAR Data

  • Sep 24, 2024
  • Sihan Chen +4
  • Research Article
  • Citations55

The Generalized Log-Ratio Transformation: Learning Shape and Adjacency Priors for Simultaneous Thigh Muscle Segmentation.

  • Feb 12, 2015
  • IEEE Transactions on Medical Imaging
  • Shawn Andrews +1
  • Book Chapter
  • Citations27

Few-Shot Single-View 3-D Object Reconstruction with Compositional Priors

  • Jan 01, 2020
  • Mateusz Michalkiewicz +5
  • PDF
  • Research Article
  • Citations3

Neural Vector Fields for Implicit Surface Representation and Inference

  • Oct 22, 2024
  • International Journal of Computer Vision
  • Edoardo Mello Rella +3
  • Preprint Article

Extracting planes from sparse 3D LiDAR data with measurement noise model

  • Nov 27, 2024
  • Linkun He +1
  • Research Article
  • Citations28

LiDAR-LLM: Exploring the Potential of Large Language Models for 3D LiDAR Understanding

  • Apr 11, 2025
  • Proceedings of the AAAI Conference on Artificial Intelligence
  • Senqiao Yang +10
  • Research Article
  • Citations1

Analysis on phase information of polarized radar in monitoring vegetation

  • Sep 01, 2001
  • Chinese Science Bulletin
  • Qingguo Feng +1
  • Research Article
  • Citations8

High-throughput 3D shape completion of potato tubers on a harvester

  • Nov 28, 2024
  • Computers and Electronics in Agriculture
  • Pieter M Blok +5
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.