mlsanityData to 14 Oct 2026

CHARM3R: Towards Unseen Camera Height Robust Monocular 3D Detector

arXiv 2508.11185 ↗First posted 15 Aug 2025cs.CV cs.LGcited by 6

Not on the map. Why?

Abstract (arXiv)

Monocular 3D object detectors, while effective on data from one ego camera height, struggle with unseen or out-of-distribution camera heights. Existing methods often rely on Plucker embeddings, image transformations or data augmentation. This paper takes a step towards this understudied problem by first investigating the impact of camera height variations on state-of-the-art (SoTA) Mono3D models. With a systematic analysis on the extended CARLA dataset with multiple camera heights, we observe that depth estimation is a primary factor influencing performance under height variations. We mathematically prove and also empirically observe consistent negative and positive trends in mean depth error of regressed and ground-based depth models, respectively, under camera height changes. To mitigate this, we propose Camera Height Robust Monocular 3D Detector (CHARM3R), which averages both depth estimates within the model. CHARM3R improves generalization to unseen camera heights by more than 45%, achieving SoTA performance on the CARLA dataset. Codes and Models at https://github.com/abhi1kumar/CHARM3R

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 92 of its 114 references that are in the map. First posted on arXiv 15 Aug 2025.

Builds on

92 of 114 references.

References · 92

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  2. 2020End-to-End Object Detection with Transformerscited by 7,414
  3. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,523
  4. 2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,247
  5. 2022BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformerscited by 1,218
  6. 2016Group Equivariant Convolutional Networkscited by 1,165
  7. 2018PointRCNN: 3D Object Proposal Generation and Detection from Point Cloudcited by 941
  8. 2017PoseCNN: A Convolutional Neural Network for 6D Object Pose Estimation in Cluttered Scenescited by 775
  9. 2021BEVDet: High-performance Multi-camera 3D Object Detection in Bird-Eye-Viewcited by 713
  10. 2018Pseudo-LiDAR from Visual Depth Estimation: Bridging the Gap in 3D Object Detection for Autonomous Drivingcited by 457
  11. 2022Last Layer Re-Training is Sufficient for Robustness to Spurious Correlationscited by 381
  12. 2022PETRv2: A Unified Framework for 3D Perception from Multi-Camera Imagescited by 352
38 more
  1. 2021Categorical Depth Distribution Network for Monocular 3D Object Detectioncited by 327
  2. 2021Is Pseudo-Lidar needed for Monocular 3D Object detection?cited by 249
  3. 2020How Neural Networks Extrapolate: From Feedforward to Graph Neural Networkscited by 222
  4. 2019Pix2Pose: Pixel-Wise Coordinate Regression of Objects for 6D Pose Estimationcited by 218
  5. 2022BEVerse: Unified Perception and Prediction in Birds-Eye-View for Vision-Centric Autonomous Drivingcited by 191
  6. 2021Objects are Different: Flexible Monocular 3D Object Detectioncited by 185
  7. 2022Time Will Tell: New Outlooks and A Baseline for Temporal Multi-View 3D Object Detectioncited by 178
  8. 2022Omni3D: A Large Benchmark and Model for 3D Object Detection in the Wildcited by 160
  9. 2021ST3D: Self-training for Unsupervised Domain Adaptation on 3D Object Detectioncited by 159
  10. 2021Geometry Uncertainty Projection Network for Monocular 3D Object Detectioncited by 156
  11. 2024VD3D: Taming Large Video Diffusion Transformers for 3D Camera Controlcited by 156
  12. 2020MonoPair: Monocular 3D Object Detection Using Pairwise Spatial Relationshipscited by 153
  13. 2020Train in Germany, Test in The USA: Making 3D Object Detectors Generalizecited by 151
  14. 2021Delving into Localization Errors for Monocular 3D Object Detectioncited by 140
  15. 2019Accurate Monocular Object Detection via Color-Embedded 3D Reconstruction for Autonomous Drivingcited by 137
  16. 2024Cameras as Rays: Pose Estimation via Ray Diffusioncited by 131
  17. 2022Vision-Centric BEV Perception: A Surveycited by 126
  18. 2017Augmented Reality Meets Computer Vision : Efficient Data Generation for Urban Driving Scenescited by 122
  19. 2020Kinematic 3D Object Detection in Monocular Videocited by 116
  20. 2022MonoDTR: Monocular 3D Object Detection with Depth-Aware Transformercited by 109
  21. 20183D-LaneNet: End-to-End 3D Multiple Lane Detectioncited by 107
  22. 2017Deep MANTA: A Coarse-to-fine Many-Task Network for joint 2D and 3D vehicle analysis from monocular imagecited by 95
  23. 2021SPG: Unsupervised Domain Adaptation for 3D Object Detection via Semantic Point Generationcited by 94
  24. 2022BEVStereo: Enhancing Depth Estimation in Multi-view 3D Object Detection with Dynamic Temporal Stereocited by 90
  25. 2020LUVLi Face Alignment: Estimating Landmarks' Location, Uncertainty, and Visibility Likelihoodcited by 87
  26. 2021AutoShape: Real-Time Shape-Aware Monocular 3D Object Detectioncited by 87
  27. 2021Evading the Simplicity Bias: Training a Diverse Set of Models Discovers Solutions with Superior OOD Generalizationcited by 83
  28. 2019How do neural networks see depth in single images?cited by 82
  29. 2019CAM-Convs: Camera-Aware Multi-Scale Convolutions for Single-View Depthcited by 75
  30. 2020Gen-LaneNet: A Generalized and Scalable Approach for 3D Lane Detectioncited by 75
  31. 2021GrooMeD-NMS: Grouped Mathematically Differentiable NMS for Monocular 3D Object Detectioncited by 66
  32. 20223D Object Detection from Images for Autonomous Driving: A Surveycited by 66
  33. 2023Shadows Don't Lie and Lines Can't Bend! Generative Models don't know Projective Geometry...for nowcited by 61
  34. 2022Diversify and Disambiguate: Learning From Underspecified Datacited by 56
  35. 2022STS: Surround-view Temporal Stereo for Multi-view 3D Detectioncited by 53
  36. 2023Temporal Enhanced Training of Multi-view 3D Object Detector via Historical Object Predictioncited by 49
  37. 2022ID and OOD Performance Are Sometimes Inversely Correlated on Real-world Datasetscited by 44
  38. 2023DistillBEV: Boosting Multi-Camera 3D Object Detection with Cross-Modal Knowledge Distillationcited by 41

50 most cited of 92 All 92 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 6.

6 papers.

  1. 2026FusionAgent: A Multimodal Agent with Dynamic Model Selection for Human Recognitioncited by 4
  2. 2026Towards Intrinsic-Aware Monocular 3D Object Detectioncited by 4
  3. 2026EmoTaG: Emotion-Aware Talking Head Synthesis on Gaussian Splatting with Few-Shot Personalizationcited by 3
  4. 2026DepthAgent: Towards Better Universal Depth Estimation via Sample-wise Expert Selectioncited by 1
  5. 2026OmniAct3D: Leveraging Foundation Geometry and Evidence-Grounded Reasoning for Panoramic 3D Detectioncited by 0
  6. 2026Autonomous Driving Research Requires a Community-Driven Data Paradigmcited by 0