---
license: mit
tags:
- image-classification
- pytorch
- skin-analysis
- dermatology
- computer-vision
datasets:
- custom
metrics:
- accuracy
- f1
pipeline_tag: image-classification
widget:
- src: https://example.com/dry-skin-sample.jpg
  example_title: Dry Skin
- src: https://example.com/oily-skin-sample.jpg
  example_title: Oily Skin
---

# 🔬 Skin Type Classification Model

A deep learning model for classifying skin types into **dry** and **oily** categories using computer vision.

## Model Description

This model is based on ResNet50 architecture and has been fine-tuned specifically for skin type classification. It can analyze facial skin images and determine whether the skin type is dry or oily with high accuracy.

### Key Features
- **Architecture**: ResNet50-based classification model
- **Classes**: 2 (dry, oily)
- **Input**: RGB images (224x224 pixels)
- **Framework**: PyTorch + Transformers
- **Performance**: High accuracy on skin type classification

## Intended Use

### Primary Use Cases
- Dermatological analysis and skin assessment
- Cosmetic product recommendation systems
- Skincare routine personalization
- Medical research and skin health monitoring

### Limitations
- Designed specifically for facial skin analysis
- Requires good lighting and clear skin visibility
- Not suitable for medical diagnosis (for research/cosmetic use only)
- Performance may vary across different skin tones and ethnicities

## How to Use

### Quick Start with Transformers

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

# Load model and processor
model = AutoModelForImageClassification.from_pretrained("your-username/skin-type-classifier")
processor = AutoImageProcessor.from_pretrained("your-username/skin-type-classifier")

# Load and process image
image = Image.open("path/to/skin/image.jpg")
inputs = processor(images=image, return_tensors="pt")

# Make prediction
with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)
    predicted_class = predictions.argmax().item()

# Get result
labels = ["dry", "oily"]
confidence = predictions[0][predicted_class].item()
print(f"Predicted skin type: {labels[predicted_class]} (confidence: {confidence:.2%})")
```

### Using the Pipeline API

```python
from transformers import pipeline

# Create classification pipeline
classifier = pipeline("image-classification", model="your-username/skin-type-classifier")

# Classify image
result = classifier("path/to/skin/image.jpg")
print(result)
```

## Model Details

### Architecture
- **Base Model**: ResNet50
- **Modification**: Custom classification head with 2 output classes
- **Input Size**: 224 × 224 × 3 (RGB)
- **Parameters**: ~25M parameters

### Training Details
- **Dataset**: Custom skin type classification dataset
- **Preprocessing**: 
  - Resize to 224×224 pixels
  - Normalization: ImageNet statistics
  - Data augmentation applied during training
- **Training Framework**: PyTorch
- **Optimization**: Adam optimizer with learning rate scheduling

### Performance Metrics
- **Training Accuracy**: High performance on validation set
- **Inference Speed**: Fast inference suitable for real-time applications
- **Model Size**: ~94MB

## Technical Specifications

### Input Format
- **Type**: RGB Images
- **Size**: 224 × 224 pixels
- **Format**: PIL Image, numpy array, or torch tensor
- **Normalization**: ImageNet mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]

### Output Format
- **Type**: Classification logits
- **Classes**: 
  - 0: "dry" - Dry skin type
  - 1: "oily" - Oily skin type
- **Output**: Softmax probabilities for each class

## Ethical Considerations

### Bias and Fairness
- Model trained on diverse skin types but may have limitations
- Users should be aware of potential biases in skin tone representation
- Continuous evaluation needed for fair performance across demographics

### Privacy
- Model processes images locally - no data transmission required
- Users responsible for ensuring proper consent when analyzing others' images
- Recommend anonymization of facial features when possible

## License

This model is released under the MIT License. See LICENSE file for details.

## Citation

If you use this model in your research, please cite:

```bibtex
@misc{skin-type-classifier-2025,
  title={Skin Type Classification Model},
  author={Your Name},
  year={2025},
  howpublished={\\url{https://huggingface.co/your-username/skin-type-classifier}},
}
```

## Contact

For questions, issues, or collaboration opportunities, please reach out through the Hugging Face model page or GitHub repository.

---

**Disclaimer**: This model is for research and cosmetic purposes only. It should not be used for medical diagnosis or treatment decisions. Always consult healthcare professionals for medical concerns.