LeVJEPA: Efficient & Scalable Video Pretraining without the Heuristics
[Widzenie komputerowe] Autorzy: Lukas Kuhn, Lucas Maes, Giuseppe Serra i 4 więcej
Video carries the temporal structure of the physical world, yet learning representations from it has remained computationally expensive: prevailing self-supervised methods either prevent representation collapse through architectural asymmetries, coupling an exponential-moving-average target encoder, a stop-gradient, and a capacity-limited predictor, or circumvent it by reconstructing masked content in pixel space. We introduce LeVJEPA, the first video encoder trained under LeJEPA's collapse-free objective, which dispenses with both. A single encoder is trained with an invariance loss over global and local views of a clip, regularized by SIGReg, which excludes collapse with a provable guarantee. The architecture reduces to an encoder and a projector, and the objective to a single hyperparameter. This formulation admits two properties. First, the cost of pretraining is governed by the num…
- Kategoria arxiv
- cs.CV
- Kategoria
- Widzenie komputerowe