mlsanityData to 7 Oct 2026

Mitigating Perspective Distortion-induced Shape Ambiguity in Image Crops

arXiv 2312.06594 ↗First posted 11 Dec 2023cs.CV cs.AI cs.LGcited by 4

Not on the map. Why?

Abstract (arXiv)

Objects undergo varying amounts of perspective distortion as they move across a camera's field of view. Models for predicting 3D from a single image often work with crops around the object of interest and ignore the location of the object in the camera's field of view. We note that ignoring this location information further exaggerates the inherent ambiguity in making 3D inferences from 2D images and can prevent models from even fitting to the training data. To mitigate this ambiguity, we propose Intrinsics-Aware Positional Encoding (KPE), which incorporates information about the location of crops in the image and camera intrinsics. Experiments on three popular 3D-from-a-single-image benchmarks: depth prediction on NYU, 3D object detection on KITTI & nuScenes, and predicting 3D shapes of articulated objects on ARCTIC, show the benefits of KPE.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Builds on 36 of its 47 references that are in the map. First posted on arXiv 11 Dec 2023.

Builds on

36 of 47 references.

References · 36

  1. 2017Attention Is All You Needcited by 59,932
  2. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  4. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,617
  5. 2017Mask R-CNNcited by 7,259
  6. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 5,986
  7. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,581
  8. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,485
  9. 2021BEiT: BERT Pre-Training of Image Transformerscited by 2,360
  10. 2021Vision Transformers for Dense Predictioncited by 1,789
  11. 2019Towards Robust Monocular Depth Estimation: Mixing Datasets for Zero-shot Cross-dataset Transfercited by 1,484
  12. 2015Spatial Transformer Networkscited by 1,423
24 more
  1. 2014Depth Map Prediction from a Single Image using a Multi-Scale Deep Networkcited by 1,264
  2. 2017End-to-end Recovery of Human Shape and Posecited by 883
  3. 2023ZoeDepth: Zero-shot Transfer by Combining Relative and Metric Depthcited by 646
  4. 2020GRAB: A Dataset of Whole-Body Human Grasping of Objectscited by 397
  5. 2019SinGAN: Learning a Generative Model from a Single Natural Imagecited by 348
  6. 2021PARE: Part Attention Regressor for 3D Human Body Estimationcited by 334
  7. 2018Learning Category-Specific Mesh Reconstruction from Image Collectionscited by 317
  8. 2022ARCTIC: A Dataset for Dexterous Bimanual Hand-Object Manipulationcited by 306
  9. 2018An Intriguing Failing of Convolutional Neural Networks and the CoordConv Solutioncited by 302
  10. 20193D Hand Shape and Pose from Images in the Wildcited by 190
  11. 2022Omni3D: A Large Benchmark and Model for 3D Object Detection in the Wildcited by 159
  12. 2019End-to-end Hand Mesh Recovery from a Monocular RGB Imagecited by 124
  13. 2022A Length-Extrapolatable Transformercited by 123
  14. 2021Keypoint Transformer: Solving Joint Identification in Challenging Hands and Object Interactions for Accurate 3D Pose Estimationcited by 122
  15. 2023Towards Zero-Shot Scale-Aware Monocular Depth Estimationcited by 100
  16. 2022LocalBins: Improving Depth Estimation by Learning Local Distributionscited by 92
  17. 2019CAM-Convs: Camera-Aware Multi-Scale Convolutions for Single-View Depthcited by 75
  18. 2022Perspective Fields for Single Image Camera Calibrationcited by 75
  19. 2023GTA: A Geometry-Aware Attention Mechanism for Multi-View Transformerscited by 75
  20. 2020Positional Encoding as Spatial Inductive Bias in GANscited by 69
  21. 2021Input-level Inductive Biases for 3D Reconstructioncited by 31
  22. 2015Amodal Completion and Size Constancy in Natural Scenescited by 21
  23. 2022Depth Field Networks for Generalizable Multi-view Scene Representationcited by 17
  24. 2023Improving Equivariance in State-of-the-Art Supervised Depth and Normal Predictorscited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 2; 2026: 2.

4 papers.

  1. 2025How Do I Do That? Synthesizing 3D Hand Motion and Contacts for Everyday Interactionscited by 10
  2. 2025Bimanual 3D Hand Motion and Articulation Forecasting in Everyday Imagescited by 1
  3. 2026Improving and Evaluating Hand-Object Interaction Detectioncited by 1
  4. 2026HandMvNet: Real-Time 3D Hand Pose Estimation Using Multi-View Cross-Attention Fusioncited by 0