shiqixudev / DINO_concise

Concise DINO Code:

---

Background:

I've streamlined the DinoV2 code repository and integrated a wrapper for clarity. This allows you to easily understand the core mechanics of how the model functions, without being sidetracked by the complexities of distributed computing and other extraneous add-ons that obscure the model's primary operations.

Outline:

    1. Dataloading an Augmentation
    2. Dino Model
    3. Training Methods
    4. Model Demo
    5. Pretrained Model

Publications:

‣ 2021 Emerging Properties in Self-Supervised Vision Transformers

‣ 2023 DINOv2: Learning Robust Visual Features without Supervision

Data:

Link to a sample dataset