---
language: en
license: apache-2.0
pipeline_tag: image-classification
datasets:
- surajbijjahalli/ISIC2018
base_model:
- facebook/dinov2-base
metrics:
- accuracy
widget:
  - src: https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/cats.png
---

# Skin Disease Classification using DINOv2 (ISIC2018)

This model classifies images of skin lesions into one of the predefined categories from the ISIC2018 dataset. It is fine-tuned on top of the `facebook/dinov2-base` Vision Transformer backbone for improved performance in medical image classification tasks.

---

## Model Details

- **Developed by:** Karl1hik
- **Finetuned from model:** [`facebook/dinov2-base`](https://huggingface.co/facebook/dinov2-base)
- **Dataset used:** [`ISIC2018`](https://huggingface.co/datasets/surajbijjahalli/ISIC2018)
- **Task:** Image classification (skin lesion diagnosis)
- **License:** Apache 2.0

---

## Uses

### Direct Use
This model can be used directly for classifying dermatoscopic images from the ISIC2018 dataset into one of the skin disease categories such as melanoma, nevus, basal cell carcinoma, etc.

### Intended Users
- Medical researchers
- Dermatology assistants
- ML practitioners working on medical imaging

### Out-of-Scope Use
This model should not be used as a standalone diagnostic tool. Clinical decisions should not rely solely on model predictions.

---

## How to Use

```python
from transformers import AutoImageProcessor, AutoModelForImageClassification
from PIL import Image
import torch

image = Image.open("your_skin_image.jpg")
processor = AutoImageProcessor.from_pretrained("kar1hik/computer-vision-project")
model = AutoModelForImageClassification.from_pretrained("kar1hik/computer-vision-project")

inputs = processor(images=image, return_tensors="pt")
with torch.no_grad():
    logits = model(**inputs).logits
predicted_class = logits.argmax(-1).item()