Bootstrapping Objectness from Videos by Relaxed Common Fate and Visual Grouping

😲 Segmenting objects in videos without human annotations 🤯

UC Berkeley
Microsoft Research Asia
UC Berkeley/UMich

TL;DR: Segmenting objects in videos without human annotations by Relaxed Common Fate and Visual Grouping.

Our Task

We study learning object segmentation from unlabeled videos. No human annotation is used.

Our Framework RCF

We learn an image segmenter first in the loop of approximating optical flow with constant segment flow plus small within-segment residual flow, and then by refining it for more coherent appearance and statistical figure-ground relevance.

RCF's Performance

On unsupervised video object segmentation, using only ResNet and convolutional heads, our model surpasses the state-of-the-art by absolute gains of 7/9/5% on DAVIS16 / STv2 / FBMS59 respectively, demonstrating the effectiveness of our ideas.

Citation

If you use this work or find it helpful, please consider citing:

                    @article{lian2023bootstrapping,
                        title={Bootstrapping Objectness from Videos by Relaxed Common Fate and Visual Grouping},
                        author={Lian, Long and Wu, Zhirong and Yu, Stella X},
                        journal={arXiv preprint arXiv:2304.08025},
                        year={2023}
                    }
                      
                    @article{lian2022improving,
                        title={Improving Unsupervised Video Object Segmentation with Motion-Appearance Synergy},
                        author={Lian, Long and Wu, Zhirong and Yu, Stella X},
                        journal={arXiv preprint arXiv:2212.08816},
                        year={2022}
                    }
                

Credit: The design of this project page references the project pages of NeRF, DeepMotionEditing, and LERF.