Learning Correspondence from the Cycle-Consistency of Time

@article{Wang2019LearningCF,
  title={Learning Correspondence from the Cycle-Consistency of Time},
  author={Xiaolong Wang and Allan Jabri and Alexei A. Efros},
  journal={CoRR},
  year={2019},
  volume={abs/1903.07593}
}
We introduce a self-supervised method for learning visual correspondence from unlabeled video. The main idea is to use cycle-consistency in time as free supervisory signal for learning visual representations from scratch. At training time, our model learns a feature map representation to be useful for performing cycle-consistent tracking. At test time, we use the acquired representation to find nearest neighbors across space and time. We demonstrate the generalizability of the representation… CONTINUE READING
Tweets
This paper has been referenced on Twitter 26 times. VIEW TWEETS

Figures, Tables, and Topics from this paper.

References

Publications referenced by this paper.
SHOWING 1-10 OF 92 REFERENCES

Similar Papers

Loading similar papers…