3ec10e6c76
* First draft * More improvements * Convert patch embedding layer * Convert all weights * Make conversion work * Improve conversion script * Fix style * Make all tests pass * Add image processor to auto mapping * Add swiglu ffn * Add image processor to conversion script * Fix conversion of giant model * Fix documentation * Fix style * Fix tests * Address comments * Address more comments * Remove unused arguments * Remove more arguments * Rename parameters * Include mask token * Address comments * Add docstring * Transfer checkpoints * Empty commit
3.0 KiB
DINOv2
Overview
The DINOv2 model was proposed in DINOv2: Learning Robust Visual Features without Supervision by Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy Vo, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, Mahmoud Assran, Nicolas Ballas, Wojciech Galuba, Russell Howes, Po-Yao Huang, Shang-Wen Li, Ishan Misra, Michael Rabbat, Vasu Sharma, Gabriel Synnaeve, Hu Xu, Hervé Jegou, Julien Mairal, Patrick Labatut, Armand Joulin, Piotr Bojanowski. DINOv2 is an upgrade of DINO, a self-supervised method applied on Vision Transformers. This method enables all-purpose visual features, i.e., features that work across image distributions and tasks without finetuning.
The abstract from the paper is the following:
The recent breakthroughs in natural language processing for model pretraining on large quantities of data have opened the way for similar foundation models in computer vision. These models could greatly simplify the use of images in any system by producing all-purpose visual features, i.e., features that work across image distributions and tasks without finetuning. This work shows that existing pretraining methods, especially self-supervised methods, can produce such features if trained on enough curated data from diverse sources. We revisit existing approaches and combine different techniques to scale our pretraining in terms of data and model size. Most of the technical contributions aim at accelerating and stabilizing the training at scale. In terms of data, we propose an automatic pipeline to build a dedicated, diverse, and curated image dataset instead of uncurated data, as typically done in the self-supervised literature. In terms of models, we train a ViT model (Dosovitskiy et al., 2020) with 1B parameters and distill it into a series of smaller models that surpass the best available all-purpose features, OpenCLIP (Ilharco et al., 2021) on most of the benchmarks at image and pixel levels.
Tips:
- One can use [
AutoImageProcessor] class to prepare images for the model.
This model was contributed by nielsr. The original code can be found here.
Dinov2Config
autodoc Dinov2Config
Dinov2Model
autodoc Dinov2Model - forward
Dinov2ForImageClassification
autodoc Dinov2ForImageClassification - forward