Predicting future dynamics is crucial for applications like autonomous driving and robotics, where understanding the environment is key. Existing pixel-level methods are computationally expensive and often focus on irrelevant details. To address these challenges, we introduce DINO-Foresight, a novel framework that operates in the semantic feature space of pretrained Vision Foundation Models (VFMs). Our approach trains a masked feature transformer in a self-supervised manner to predict the evolution of VFM features over time. By forecasting these features, we can apply off-the-shelf, task-specific heads for various scene understanding tasks. In this framework, VFM features are treated as a latent space, to which different heads attach to perform specific tasks for future-frame analysis. Extensive experiments show the very strong performance, robustness and scalability of our framework. Project page and code at https://dino-foresight.github.io/ .
@article{arxiv.2412.11673,
title = {DINO-Foresight: Looking into the Future with DINO},
author = {Efstathios Karypidis and Ioannis Kakogeorgiou and Spyros Gidaris and Nikos Komodakis},
journal= {arXiv preprint arXiv:2412.11673},
year = {2025}
}