File size: 1,844 Bytes
66ab751
 
 
 
 
 
 
 
 
 
 
 
 
 
0b2e772
66ab751
 
 
0b2e772
66ab751
0b2e772
66ab751
 
 
 
 
 
30e19e0
66ab751
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
---
license: cc-by-nc-4.0
language:
- en
base_model:
- timm/vit_large_patch14_dinov2.lvd142m
pipeline_tag: image-classification
tags:
- Fluorescein Angiography (FA)
- multiclass classification
---
# DINOv2-Large (HyperF_Type)

Classification of hyperfluorescence type on Fluorescein Angiography (FA) images.

Multiclass problem: 0 (leakage), 1 (staining), 2 (no), 3 (pooling), 4 (window defect)
       
This model is obtained by finetuning the pretrained architecture of HuggingFace ``vit_large_patch14_dinov2.lvd142m``.

Trained on: [AngioReport](https://pubmed.ncbi.nlm.nih.gov/40610046/)

## Usage

Check instruction at the following [repository](https://gitlab.idiap.ch/medai/software/paper/fm-overspecialization)

## Model output structure

`[batch_size, num_classes]` (`num_classes=5`)

## Related publication(s)

```
@article{xu_angioreport_2025,
        title = {{AngioReport}: dataset and baseline methods for fundus angiography report generation},
        issn = {0007-1161},
        url = {https://bjo.bmj.com/content/early/2025/07/02/bjo-2024-327006},
        doi = {10.1136/bjo-2024-327006},
        journal = {British Journal of Ophthalmology},
        author = {Xu, Pusheng and Chotcomwongse, Peranut and Zhang, Weiyi and Chen, Xiaolan and Wu, Xinyuan and Chung, Florence H T and Zhang, Xueli and He, Mingguang and Shi, Danli and Ruamviboonsuk, Paisan},
        year = {2025},
}
@misc{he2021maskedautoencodersscalablevision,
        title={Masked Autoencoders Are Scalable Vision Learners},
        author={Kaiming He and Xinlei Chen and Saining Xie and Yanghao Li and Piotr Dollár and Ross Girshick},
        year={2021},
        eprint={2111.06377},
        archivePrefix={arXiv},
        primaryClass={cs.CV},
        url={https://arxiv.org/abs/2111.06377},
}
```

## Author(s)

* Roberto Pulvirenti, Idiap Research Institute