DINO-MX: A Modular & Flexible Framework for Self-Supervised Learning

TL;DR

DINO-MX is a flexible, modular framework for self-supervised vision learning that supports diverse architectures and training strategies, achieving competitive results with reduced computational costs and enhanced interpretability.

Contribution

It introduces a unified, extensible training system combining multiple DINO variants, supporting various architectures, strategies, and data types, with tools for interpretability and data augmentation.

Findings

Achieves competitive performance on diverse datasets.

Reduces computational costs significantly.

Enhances interpretability and localization without extra detection heads.

Abstract

Vision Foundation Models (VFMs) have advanced representation learning through self-supervised methods. However, existing training pipelines are often inflexible, domain-specific, or computationally expensive, which limits their usability across different domains and resource settings. DINO-MX is a modular and extensible training framework that combines the core principles of DINO, DINOv2 and DINOv3 within a unified configuration-driven system. It supports a variety of transformer-based architectures and is fully compatible with the Hugging Face ecosystem. The framework includes multiple training strategies such as low-rank adaptation (LoRA), layer freezing, and knowledge distillation, along with support for distributed training through both Distributed Data Parallel (DDP) and Fully Sharded Data Parallel (FSDP). DINO-MX is designed to work with both natural and specialized data types, including single- and multi-channel images. Experimental results on diverse datasets show that DINO-MX achieves competitive performance while significantly reducing computational costs. Additionally, it offers interpretability tools and a label-guided data augmentation method that improves attention-based localization without the need for extra detection or segmentation heads. DINO-MX provides a reproducible and scalable foundation for developing, adapting, and benchmarking self-supervised vision models across a range of research and real-world applications.