FastDINOv2: Frequency Based Curriculum Learning Improves Robustness and Training Speed

Large-scale vision foundation models such as DINOv2 boast impressive performances by leveraging massive architectures and training datasets. But numerous scenarios require practitioners to reproduce those pre-training solutions, such as on private data, new modalities, or simply for scientific questioning-which is currently extremely demanding computation-wise. We thus propose a novel pretraining strategy for DINOv2 that simultaneously accelerates convergence-and strengthens robustness to common corruptions as a by-product. Our approach involves a frequency filtering curriculum-low-frequency being seen first-and the Gaussian noise patching augmentation. Applied to a ViT-B/16 backbone trained on ImageNet-1K, while pre-training time and FLOPs are reduced by 1.6× and 2.25×, our method still achieves matching robustness in corruption benchmarks (ImageNet-C) and maintains competitive linear probing performance compared with baseline. This dual benefit of efficiency and robustness makes large-scale self-supervised foundation modeling more attainable, while opening the door to novel exploration around data curriculum and augmentation as means to improve self-supervised learning models robustness. The code is available at https://github.com/KevinZ0217/fast_dinov2 Figure 1: The FastDINOv2 training pipeline comprises two stages. In the first stage, the initial 75% of training epochs utilize only low-frequency features extracted via downsampling. In the second stage, the remaining 25% of epochs employ full-resolution images with Gaussian noise patching.

Table 1: Comparison of training costs, evaluation on ImageNet-C, and linear probing on the ImageNet-1K validation set using a frozen ViT-B DINOv2 backbone trained on ImageNet-1K. Training Method Training Time (days)(↓) ImageNet-1K(↑) ImageNet-C (↓) GFLOPs (↓) DINOv2 16.64 (NVIDIA L40S) 77.8% 56.5% 493.76 FastDINOv2 10.32 (NVIDIA L40S) 76.2% 56.7% 219.92 39th Conference on Neural Information Processing Systems (NeurIPS 2025).