mlsanityData to 14 Oct 2026

Deep Feedback Inverse Problem Solver

arXiv 2101.07719 ↗First posted 19 Jan 2021cs.CV cs.ROcited by 11

Not on the map. Why?

Abstract (arXiv)

We present an efficient, effective, and generic approach towards solving inverse problems. The key idea is to leverage the feedback signal provided by the forward process and learn an iterative update model. Specifically, at each iteration, the neural network takes the feedback as input and outputs an update on the current estimation. Our approach does not have any restrictions on the forward process; it does not require any prior knowledge either. Through the feedback information, our model not only can produce accurate estimations that are coherent to the input observation but also is capable of recovering from early incorrect predictions. We verify the performance of our approach over a wide range of inverse problems, including 6-DOF pose estimation, illumination estimation, as well as inverse kinematics. Comparing to traditional optimization-based methods, we can achieve comparable or better performance while being two to three orders of magnitude faster. Compared to deep learning-based approaches, our model consistently improves the performance on all metrics. Please refer to the project page for videos, animations, supplementary materials, etc.

Key facts. Cited by 11 papers in the map (data through 14 Oct 2026). Builds on 45 of its 80 references that are in the map. First posted on arXiv 19 Jan 2021.

Builds on

45 of 80 references.

References · 45

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  2. 2015ShapeNet: An Information-Rich 3D Model Repositorycited by 2,897
  3. 2016Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Networkcited by 1,908
  4. 2016Context Encoders: Feature Learning by Inpaintingcited by 1,428
  5. 2018ESRGAN: Enhanced Super-Resolution Generative Adversarial Networkscited by 1,254
  6. 2018On the Continuity of Rotation Representations in Neural Networkscited by 1,214
  7. 2017End-to-end Recovery of Human Shape and Posecited by 885
  8. 2016Deep Multi-scale Convolutional Neural Network for Dynamic Scene Deblurringcited by 727
  9. 2018Image Inpainting for Irregular Holes Using Partial Convolutionscited by 505
  10. 2017Deep Laplacian Pyramid Networks for Fast and Accurate Super-Resolutioncited by 453
  11. 2017Neural 3D Mesh Renderercited by 437
  12. 2019Soft Rasterizer: A Differentiable Renderer for Image-based 3D Reasoningcited by 420
33 more
  1. 2019SynSin: End-to-end View Synthesis from a Single Imagecited by 364
  2. 2017On human motion prediction using recurrent neural networkscited by 338
  3. 2016Convolutional Pose Machinescited by 325
  4. 2018Learning Category-Specific Mesh Reconstruction from Image Collectionscited by 317
  5. 2013DeepPose: Human Pose Estimation via Deep Neural Networkscited by 305
  6. 2017"Zero-Shot" Super-Resolution using Deep Internal Learningcited by 286
  7. 2018DeepIM: Deep Iterative Matching for 6D Pose Estimationcited by 283
  8. 2019DeepView: View Synthesis with Learned Gradient Descentcited by 253
  9. 2017Learning Efficient Point Cloud Generation for Dense 3D Object Reconstructioncited by 133
  10. 2015Human Pose Estimation with Iterative Error Feedbackcited by 121
  11. 2018QuaterNet: A Quaternion-based Recurrent Model for Human Motioncited by 109
  12. 2018Single Image Reflection Separation with Perceptual Lossescited by 98
  13. 2017Self-supervised Learning of Motion Capturecited by 97
  14. 2018ST-GAN: Spatial Transformer Generative Adversarial Networks for Image Compositingcited by 90
  15. 2018Neural Kinematic Networks for Unsupervised Motion Retargettingcited by 88
  16. 2019Oops! Predicting Unintentional Action in Videocited by 82
  17. 2018Learning Intrinsic Image Decomposition from Watching the Worldcited by 74
  18. 2019Rendering Synthetic Objects into Legacy Photographscited by 72
  19. 2017One Network to Solve Them All --- Solving Linear Inverse Problems using Deep Projection Modelscited by 67
  20. 2018Deep Model-Based 6D Pose Refinement in RGBcited by 59
  21. 2019Deep Rigid Instance Scene Flowcited by 54
  22. 2016Highway and Residual Networks learn Unrolled Iterative Estimationcited by 52
  23. 2017Self-Supervised Intrinsic Image Decompositioncited by 43
  24. 2016Inverse Compositional Spatial Transformer Networkscited by 29
  25. 2016Feedback Networkscited by 29
  26. 2019Soft Rasterizer: Differentiable Rendering for Unsupervised Single-View Mesh Reconstructioncited by 28
  27. 2019Photometric Mesh Optimization for Video-Aligned 3D Object Reconstructioncited by 28
  28. 2017Adversarial Inverse Graphics Networks: Learning 2D-to-3D Lifting and Image-to-Image Translation from Unpaired Supervisioncited by 27
  29. 20183D-Aware Scene Manipulation via Inverse Graphicscited by 24
  30. 2016Training a Feedback Loop for Hand Pose Estimationcited by 21
  31. 2016Chained Predictions Using Convolutional Neural Networkscited by 15
  32. 2015Learning Super-Resolution Jointly from External and Internal Examplescited by 2
  33. 2018Learning Direct Optimization for Scene Understandingcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 4; 2022: 0; 2023: 3; 2024: 1; 2025: 2; 2026: 1.

11 papers.

  1. 2021Back to the Feature: Learning Robust Camera Localization from Pixels to Posecited by 199
  2. 2023Grad-PU: Arbitrary-Scale Point Cloud Upsampling via Gradient Descent with Learned Distance Functionscited by 43
  3. 2024G3R: Gradient Guided Generalizable Reconstructioncited by 31
  4. 2021Robust Inverse Framework using Knowledge-guided Self-Supervised Learning: An application to Hydrologycited by 15
  5. 2025ReSplat: Learning Recurrent Gaussian Splattingcited by 15
  6. 2021MetaPose: Fast 3D Pose from Multiple Views without 3D Supervisioncited by 12
  7. 2025LIFe-GoM: Generalizable Human Rendering with Learned Iterative Feedback Over Multi-Resolution Gaussians-on-Meshcited by 11
  8. 2023Rapid Network Adaptation: Learning to Adapt Neural Networks Using Test-Time Feedbackcited by 10
  9. 2023TransPoser: Transformer as an Optimizer for Joint Object Shape and Pose Estimationcited by 2
  10. 2021Solving inverse problems with deep neural networks driven by sparse signal decomposition in a physics-based dictionarycited by 1
  11. 2026FlowBender: Feedback-Aware Training for Self-Correcting Conditional Flowscited by 0