🚧 This project is under active development. Model retraining and evaluation are ongoing. This project is released under the MIT License for academic and research purposes.
Coded by Ares Coding
An Android-based application that detects alcohol intoxication from facial images using image processing and machine learning techniques.
This project analyzes facial cues to classify individuals as Normal or Drunk.
🎓 Developed as part of an academic research / thesis project.
Alcohol intoxication affects facial expressions, eye behavior, and muscle control.
This application captures facial images and processes them using trained machine learning models to determine intoxication status.
The main objective of this project is to explore the feasibility of detecting alcohol intoxication through facial image analysis and deploy the solution in a real Android environment.
- 📸 Capture facial images using an Android device camera
- 🧠 Image processing for facial feature analysis
- 📊 Classification result: Normal or Drunk
- 📱 Successfully tested on a real Android device
- 🧪 Experimental comparison of multiple ML models
The system was trained and evaluated using the following models:
- CNN (Convolutional Neural Network)
- SVM (Support Vector Machine)
- Hybrid CNN–SVM Model
The hybrid approach uses CNN for feature extraction and SVM for final classification to improve performance.
- Facial images categorized into:
- Normal
- Drunk
- Dataset characteristics:
- Multiple individuals
- Various facial expressions
- Different lighting conditions
⚠️ The dataset is not publicly available due to academic and ethical considerations.
- Android (Java)
- Gradle
- Python
- OpenCV
- CNN
- SVM
- Scikit-learn
- TensorFlow
- NumPy
- Pandas
- Successfully deployed and tested on Android Emulator and real Android device
- Model retraining is currently in progress using an expanded dataset of 20k+ images of Drunk,and 20k+ Normal images.