mlsanityData to 13 Oct 2026

A Compacted Structure for Cross-domain learning on Monocular Depth and Flow Estimation

arXiv 2208.11993 ↗First posted 25 Aug 2022cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Accurate motion and depth recovery is important for many robot vision tasks including autonomous driving. Most previous studies have achieved cooperative multi-task interaction via either pre-defined loss functions or cross-domain prediction. This paper presents a multi-task scheme that achieves mutual assistance by means of our Flow to Depth (F2D), Depth to Flow (D2F), and Exponential Moving Average (EMA). F2D and D2F mechanisms enable multi-scale information integration between optical flow and depth domain based on differentiable shallow nets. A dual-head mechanism is used to predict optical flow for rigid and non-rigid motion based on a divide-and-conquer manner, which significantly improves the optical flow estimation performance. Furthermore, to make the prediction more robust and stable, EMA is used for our multi-task training. Experimental results on KITTI datasets show that our multi-task scheme outperforms other multi-task schemes and provide marked improvements on the prediction results.

Key facts. No citing papers in the map yet (data through 13 Oct 2026). Builds on 31 of its 49 references that are in the map. First posted on arXiv 25 Aug 2022.

Builds on

31 of 49 references.

References · 31

  1. 2017Attention Is All You Needcited by 60,106
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,304
  3. 2014Adam: A Method for Stochastic Optimizationcited by 37,093
  4. 2020RAFT: Recurrent All-Pairs Field Transforms for Optical Flowcited by 2,395
  5. 2014Depth Map Prediction from a Single Image using a Multi-Scale Deep Networkcited by 1,269
  6. 2017PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volumecited by 967
  7. 2016Unsupervised Monocular Depth Estimation with Left-Right Consistencycited by 819
  8. 2016FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networkscited by 782
  9. 2017Unsupervised Learning of Depth and Ego-Motion from Videocited by 779
  10. 2014Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-Scale Convolutional Architecturecited by 483
  11. 2016Deeper Depth Prediction with Fully Convolutional Residual Networkscited by 403
  12. 2016Unsupervised CNN for Single View Depth Estimation: Geometry to the Rescuecited by 368
19 more
  1. 2018GeoNet: Unsupervised Learning of Dense Depth, Optical Flow and Camera Posecited by 342
  2. 2018Competitive Collaboration: Joint Unsupervised Learning of Depth, Camera Motion, Optical Flow and Motion Segmentationcited by 222
  3. 2017UnFlow: Unsupervised Learning of Optical Flow with a Bidirectional Census Losscited by 202
  4. 2017Computer Vision for Autonomous Vehicles: Problems, Datasets and State of the Artcited by 182
  5. 2016View Synthesis by Appearance Flowcited by 172
  6. 2018DF-Net: Unsupervised Joint Learning of Depth and Flow using Cross-Task Consistencycited by 148
  7. 2019Self-supervised Learning with Geometric Constraints in Monocular Video: Connecting Flow, Depth, and Cameracited by 130
  8. 2017Occlusion Aware Unsupervised Learning of Optical Flowcited by 101
  9. 2018Deep Virtual Stereo Odometry: Leveraging Deep Depth Prediction for Monocular Direct Sparse Odometrycited by 85
  10. 2020Towards Better Generalization: Joint Depth-Pose Learning without PoseNetcited by 85
  11. 2020Self-Supervised Monocular Scene Flow Estimationcited by 64
  12. 2016Full Flow: Optical Flow Estimation By Global Optimization over Regular Gridscited by 34
  13. 2022Learning Optical Flow, Depth, and Scene Flow without Real-World Labelscited by 31
  14. 2017Monocular Depth Estimation with Hierarchical Fusion of Dilated CNNs and Soft-Weighted-Sum Inferencecited by 24
  15. 2019Video Depth Estimation by Fusing Flow-to-Depth Proposalscited by 23
  16. 2020Unsupervised Learning of Depth, Optical Flow and Pose with Occlusion from 3D Geometrycited by 21
  17. 2021Self-Supervised Generative Adversarial Network for Depth Estimation in Laparoscopic Imagescited by 21
  18. 2022Self-Supervised Depth Estimation in Laparoscopic Image using 3D Geometric Consistencycited by 19
  19. 2021H-Net: Unsupervised Attention-based Stereo Depth Estimation Leveraging Epipolar Geometrycited by 12

Built on by

Not cited yet.