Automatic acne detection and classification on facial images using Convolutional Neural Networks (CNN) and Deep Learning technology. This project aims to assist early skin analysis and can be used as the foundation for AI-powered mobile or web applications.
This project builds a CNN model trained on annotated facial image datasets to classify skin conditions, particularly acne. The deep learning model is expected to recognize acne of various severity levels from facial images. This solution can support dermatologists, skincare applications, and general users in better understanding their skin conditions.
-
🔄 Image Preprocessing & Augmentation
Includes image resizing, normalization, and augmentation to improve model generalization and robustness. -
🧠 Custom CNN Architecture
Built with TensorFlow/Keras, the CNN model extracts facial features relevant to acne detection. -
📊 Training & Validation Metrics
Visualizes model performance using accuracy/loss graphs and confusion matrices. -
📱 Deployment Potential
Ready for integration into mobile/web apps as an AI-based acne detection tool.
git clone https://github.com/username/acne-detector.git
cd acne-detectorEnsure you're using Python 3.7+, then install the required packages:
pip install -r requirements.txtTo run the model or deploy it via an API:
python main.py- Validation Accuracy: ~82%
- Training Accuracy: ~90%
- Final Validation Loss: ~3.1
- Final Training Loss: ~2.8
- F1-Score: ~0.89 (based on additional evaluation)
- The confusion matrix shows the model performs well in distinguishing between acne and non-acne facial skin conditions.
We welcome contributions from the community!
How to Contribute:
- Fork this repository
- Create a feature branch (git checkout -b your-feature)
- Commit your changes (git commit -m "Add: Your feature")
- Push to GitHub (git push origin your-feature)
- Submit a Pull Request
- Aditya Putra Sholahuddin (a.k.a Kreess)
- GitHub: @KREESS
- Email: adityasholahuddin@gmail.com
- Instagram: @adityasholahuddin