Self-Supervised Any-Point Tracking by Contrastive Random Walks
This work addresses the problem of precise point tracking in videos for computer vision applications, presenting an incremental improvement over existing self-supervised methods.
The paper tackles the Tracking Any Point (TAP) problem by introducing a self-supervised approach using a global matching transformer trained with contrastive random walks, achieving strong performance on TapVid benchmarks and outperforming previous self-supervised methods like DIFT while being competitive with supervised ones.
We present a simple, self-supervised approach to the Tracking Any Point (TAP) problem. We train a global matching transformer to find cycle consistent tracks through video via contrastive random walks, using the transformer's attention-based global matching to define the transition matrices for a random walk on a space-time graph. The ability to perform "all pairs" comparisons between points allows the model to obtain high spatial precision and to obtain a strong contrastive learning signal, while avoiding many of the complexities of recent approaches (such as coarse-to-fine matching). To do this, we propose a number of design decisions that allow global matching architectures to be trained through self-supervision using cycle consistency. For example, we identify that transformer-based methods are sensitive to shortcut solutions, and propose a data augmentation scheme to address them. Our method achieves strong performance on the TapVid benchmarks, outperforming previous self-supervised tracking methods, such as DIFT, and is competitive with several supervised methods.