mlsanityData to 7 Oct 2026

KECOR: Kernel Coding Rate Maximization for Active 3D Object Detection

arXiv 2307.07942 ↗First posted 16 Jul 2023cs.CV cs.AI cs.LGcited by 13

Not on the map. Why?

Abstract (arXiv)

Achieving a reliable LiDAR-based object detector in autonomous driving is paramount, but its success hinges on obtaining large amounts of precise 3D annotations. Active learning (AL) seeks to mitigate the annotation burden through algorithms that use fewer labels and can attain performance comparable to fully supervised learning. Although AL has shown promise, current approaches prioritize the selection of unlabeled point clouds with high uncertainty and/or diversity, leading to the selection of more instances for labeling and reduced computational efficiency. In this paper, we resort to a novel kernel coding rate maximization (KECOR) strategy which aims to identify the most informative point clouds to acquire labels through the lens of information theory. Greedy search is applied to seek desired point clouds that can maximize the minimal number of bits required to encode the latent features. To determine the uniqueness and informativeness of the selected samples from the model perspective, we construct a proxy network of the 3D detector head and compute the outer product of Jacobians from all proxy layers to form the empirical neural tangent kernel (NTK) matrix. To accommodate both one-stage (i.e., SECOND) and two-stage detectors (i.e., PVRCNN), we further incorporate the classification entropy maximization and well trade-off between detection performance and the total number of bounding boxes selected for annotation. Extensive experiments conducted on two 3D benchmarks and a 2D detection dataset evidence the superiority and versatility of the proposed approach. Our results show that approximately 44% box-level annotation costs and 26% computational time are reduced compared to the state-of-the-art AL method, without compromising detection performance.

Key facts. Cited by 13 papers in the map (data through 7 Oct 2026). Builds on 43 of its 78 references that are in the map. First posted on arXiv 16 Jul 2023.

Builds on

43 of 78 references.

References · 43

  1. 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,416
  2. 2015SSD: Single Shot MultiBox Detectorcited by 3,125
  3. 2018Neural Tangent Kernel: Convergence and Generalization in Neural Networkscited by 2,597
  4. 2019Scalability in Perception for Autonomous Driving: Waymo Open Datasetcited by 2,436
  5. 2018PointPillars: Fast Encoders for Object Detection from Point Cloudscited by 1,707
  6. 2018PointRCNN: 3D Object Proposal Generation and Detection from Point Cloudcited by 939
  7. 2019PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detectioncited by 890
  8. 2019Wide Neural Networks of Any Depth Evolve as Linear Models Under Gradient Descentcited by 732
  9. 2019Deep Batch Active Learning by Diverse, Uncertain Gradient Lower Boundscited by 602
  10. 2020A Survey of Deep Active Learningcited by 547
  11. 20203DSSD: Point-based 3D Single Stage Object Detectorcited by 440
  12. 2019BatchBALD: Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learningcited by 428
31 more
  1. 2020Voxel R-CNN: Towards High Performance Voxel-based 3D Object Detectioncited by 409
  2. 2019PIXOR: Real-time 3D Object Detection from Point Cloudscited by 358
  3. 2019Learning Loss for Active Learningcited by 335
  4. 2019STD: Sparse-to-Dense 3D Object Detector for Point Cloudcited by 282
  5. 2020Learning Diverse and Discriminative Representations via the Principle of Maximal Coding Rate Reductioncited by 141
  6. 2020Contextual Diversity for Active Learningcited by 139
  7. 2021Batch Active Learning at Scalecited by 127
  8. 2022Voxel Set Transformer: A Set-to-Set Approach to 3D Object Detection from Point Cloudscited by 102
  9. 2022PillarNet: Real-Time and High-Performance Pillar-based 3D Object Detectioncited by 99
  10. 2021Gone Fishing: Neural Active Learning with Fisher Embeddingscited by 88
  11. 2019Harnessing the Power of Infinitely Wide Deep Nets on Small-data Taskscited by 87
  12. 2020Task-Aware Variational Adversarial Active Learningcited by 84
  13. 2021Influence Selection for Active Learningcited by 82
  14. 2021Active Learning for Deep Object Detection via Probabilistic Modelingcited by 79
  15. 2021Multiple instance active learning for object detectioncited by 77
  16. 2019BayesOD: A Bayesian Approach for Uncertainty Estimation in Deep Object Detectorscited by 68
  17. 2022Active Learning by Feature Mixingcited by 63
  18. 2022Fast Finite Width Neural Tangent Kernelcited by 62
  19. 2022Entropy-based Active Learning for Object Detection with Progressive Diversity Constraintcited by 53
  20. 2018Localization-Aware Active Learning for Object Detectioncited by 48
  21. 2022VISTA: Boosting 3D Object Detection via Dual Cross-VIew SpaTial Attentioncited by 48
  22. 2022A Framework and Benchmark for Deep Batch Active Learning for Regressioncited by 47
  23. 2019Deep Active Learning for Efficient Training of a LiDAR 3D Object Detectorcited by 45
  24. 2021Frustratingly Easy Transferability Estimationcited by 44
  25. 2022Self-Supervised Learning via Maximum Entropy Codingcited by 33
  26. 2021Consistency-based Active Learning for Object Detectioncited by 32
  27. 2023Exploring Active 3D Object Detection from a Generalization Perspectivecited by 26
  28. 2022Making Look-Ahead Active Learning Strategies Feasible with Neural Tangent Kernelscited by 25
  29. 2015Hierarchical Subquery Evaluation for Active Learning on a Graphcited by 21
  30. 2022OccAM's Laser: Occlusion-based Attribution Maps for 3D Object Detectors on LiDAR Datacited by 13
  31. 2020PAL : Pretext-based Active Learningcited by 9

Built on by

Cited by year

Citing papers per year: 2023: 1; 2024: 6; 2025: 6.

13 papers.

Sort:

By area:

  1. 2024Find n' Propagate: Open-Vocabulary 3D Object Detection in Urban Environmentscited by 21
  2. 2024The Why, When, and How to Use Active Learning in Large-Data-Driven 3D Object Detection for Safe Autonomous Driving: An Empirical Explorationcited by 17
  3. 2024ActiveAnno3D -- An Active Learning Framework for Multi-Modal 3D Object Detectioncited by 17
  4. 2025Uncertainty Meets Diversity: A Comprehensive Active Learning Framework for Indoor 3D Object Detectioncited by 7
  5. 2024MOS: Model Synergy for Test-Time Adaptation on LiDAR-Based 3D Object Detectioncited by 6
  6. 2025Generalizable Prompt Learning of CLIP: A Brief Overviewcited by 3
  7. 2025IDEAL-M3D: Instance Diversity-Enriched Active Learning for Monocular 3D Detectioncited by 2
  8. 2024Distribution Discrepancy and Feature Heterogeneity for Active 3D Object Detectioncited by 1
  9. 2024STONE: A Submodular Optimization Framework for Active 3D Object Detectioncited by 1
  10. 2025SCORE: Soft Label Compression-Centric Dataset Condensation via Coding Rate Optimizationcited by 1
  11. 2025CodeMerge: Codebook-Guided Model Merging for Robust Test-Time Adaptation in Autonomous Drivingcited by 1
  12. 2023Open-CRB: Towards Open World Active Learning for 3D Object Detectioncited by 0
1 more
  1. 2025HeAL3D: Heuristical-enhanced Active Learning for 3D Object Detectioncited by 0