DDFlow: Learning Optical Flow with Unlabeled Data Distillation

By Pengpeng Liu, Irwin King, Michael R. Lyu, Jia Xu. 2019

199 citations recorded by the source corpus.

We present DDFlow, a data distillation approach to learning optical flow estimation from unlabeled data. The approach distills reliable predictions from a teacher network, and uses these predictions as annotations to guide a student network to learn optical flow. Unlike existing work relying on handcrafted energy terms to handle occlusion, our approach is data-driven, and learns optical flow for occluded pixels. This enables us to train our model with a much simpler loss function, and achieve a much higher accuracy. We conduct a rigorous evaluation on the challenging Flying Chairs, MPI Sintel, KITTI 2012 and 2015 benchmarks, and show that our approach significantly outperforms all existing unsupervised learning methods, while running at real time.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Dacheng Tao
2. Michael R. Lyu
3. Irwin King
4. Jianping Gou
5. S. Maybank
6. Baosheng Yu
7. Jia Xu
8. Pengpeng Liu

Citations in the corpus

• Knowledge Distillation: A Survey (2020, methodology): “Furthermore, the parameters of the teacher model (or the connections between layers) also contain another knowledge (Liu et al. 2019c).”
• SelFlow: Self-Supervised Learning of Optical Flow (2019, background): “40 [26] to 6.”
• CoTracker3: Simpler and Better Point Tracking by Pseudo-Labeling Real Videos (2025, unknown): “Alternatively, one can use an existing tracker to train another in a process akin to distillation [22].”