Open-source Model for Diabetic_RetinoPathy_detection - Accurately Detect Diabetic Retinopathy with 96.8% Accuracy

Diabetic RetinoPathy Detection

Developed by AsmaaElnagger

An image classification model fine-tuned based on Facebook's DINOv2 base model, specifically designed for diabetic retinopathy detection, achieving 96.8% accuracy on the evaluation set.

Image Classification

Transformers

Supports Multiple LanguagesOpen Source License:Gpl-3.0 #Diabetic Retinopathy Detection #Medical Image Classification #High-Precision Fundus Analysis

Downloads 1,008

Release Time : 3/17/2025

Model Overview

This model is a vision Transformer based on the DINOv2 architecture, fine-tuned specifically for medical image classification tasks, particularly the detection of diabetic retinopathy.

Model Features

High-Precision Medical Image Classification

Achieves 96.8% accuracy in diabetic retinopathy detection tasks

Powerful Visual Feature Extraction Based on DINOv2

Utilizes the self-supervised pre-trained DINOv2 model as a foundation, offering excellent visual representation capabilities

Multilingual Support

Supports usage in both English and Arabic environments

Model Capabilities

Medical Image Analysis

Diabetic Retinopathy Detection

Image Classification

Use Cases

Medical Diagnosis

Diabetic Retinopathy Screening

Automatically detects signs of diabetic retinopathy in fundus images

Evaluation set accuracy: 96.8%, F1 score: 0.9678

🚀 Diabetic_RetinoPathy_detection

This model is a fine - tuned version of [facebook/dinov2 - base](https://huggingface.co/facebook/dinov2 - base) for diabetic retinopathy detection, achieving high accuracy on the imagefolder dataset.

🚀 Quick Start

This model is a fine - tuned version of [facebook/dinov2 - base](https://huggingface.co/facebook/dinov2 - base) on the imagefolder dataset. It achieves the following results on the evaluation set:

Loss: 0.2262
Accuracy: 0.968
F1: 0.9678

📚 Documentation

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

🔧 Technical Details

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2e - 05
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.3853	1.0	250	0.4918	0.874	0.8729
0.5345	2.0	500	0.4390	0.878	0.8771
0.4693	3.0	750	0.3857	0.88	0.8796
0.1933	4.0	1000	0.3444	0.894	0.8948
0.3146	5.0	1250	0.2456	0.936	0.9362
0.1832	6.0	1500	0.3369	0.924	0.9229
0.1407	7.0	1750	0.3425	0.946	0.9454
0.1462	8.0	2000	0.2864	0.948	0.9476
0.0905	9.0	2250	0.2177	0.956	0.9560
0.0859	10.0	2500	0.2262	0.968	0.9678

Framework versions

Transformers 4.45.1
Pytorch 2.4.0
Datasets 3.0.1
Tokenizers 0.20.0

📄 License

This project is licensed under the GPL - 3.0 license.

📦 Model Information

Property	Details
Library Name	transformers
Base Model	facebook/dinov2 - base
Tags	generated_from_trainer
Datasets	imagefolder
Metrics	accuracy, f1
Model Name	dinov2 - base - finetuned - dia_eye
Task	Image Classification
Evaluation Dataset	imagefolder (train split)
Accuracy	0.88
F1	0.8802333980766498
Supported Languages	en, ar

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご