DINO-Tracker: Taming DINO for Self-Supervised Point Tracking in a Single Video
This work addresses video tracking challenges for computer vision applications, offering an incremental improvement by integrating existing DINO features with test-time training.
The paper tackles the problem of long-term dense tracking in video by introducing DINO-Tracker, a framework that combines test-time training with pre-trained DINO-ViT features, achieving state-of-the-art results on benchmarks and outperforming self-supervised methods while being competitive with supervised trackers.
We present DINO-Tracker -- a new framework for long-term dense tracking in video. The pillar of our approach is combining test-time training on a single video, with the powerful localized semantic features learned by a pre-trained DINO-ViT model. Specifically, our framework simultaneously adopts DINO's features to fit to the motion observations of the test video, while training a tracker that directly leverages the refined features. The entire framework is trained end-to-end using a combination of self-supervised losses, and regularization that allows us to retain and benefit from DINO's semantic prior. Extensive evaluation demonstrates that our method achieves state-of-the-art results on known benchmarks. DINO-tracker significantly outperforms self-supervised methods and is competitive with state-of-the-art supervised trackers, while outperforming them in challenging cases of tracking under long-term occlusions.