Mental Health Application Based Face Recognition

Key Features:

• Emotion Detection: Analyzes user’s facial expressions to identify emotions.
• Personalized Recommendations: Provides suggestions based on the user’s emotional state.
• Professional Integration: Notifications to contact mental health professionals if needed.

Formulation

1. Hash Function (djb2Hash)

The hash function is used to generate a unique FaceID based on facial landmarks. The formula for the hash function is:

$\text{hash}=5381\text{for each character }i\text{ in the string:}\text{hash}=(\text{hash}\times 33)+\text{charCodeAt}(i)\text{return hash }>>0\text{(unsigned 32-bit integer)}\backslash text\{hash\}\; =\; 5381\; \backslash \backslash \; \backslash text\{for\; each\; character\; \}\; i\; \backslash text\{\; in\; the\; string:\}\; \backslash \backslash \; \backslash text\{hash\}\; =\; (\backslash text\{hash\}\; \backslash times\; 33)\; +\; \backslash text\{charCodeAt\}(i)\; \backslash \backslash \; \backslash text\{return\; hash\; \}\; >>\; 0\; \backslash text\{(unsigned\; 32-bit\; integer)\}$

2. Lip Stretch Calculation (Happiness)

The lip stretch is calculated using the Euclidean distance between the left and right lip corners:

$\text{lipStretch}=\sqrt{(x_2-x_1{)}^2+(y_2-y_1{)}^2}\backslash text\{lipStretch\}\; =\; \backslash sqrt\{(x\_2\; -\; x\_1)^2\; +\; (y\_2\; -\; y\_1)^2\}$

3. Cheek Raise Calculation (Happiness)

The cheek raise is calculated as the vertical distance between the cheek and eye landmarks:

$\text{cheekRaise}=y_{\text{eye}}-y_{\text{cheek}}\backslash text\{cheekRaise\}\; =\; y\_\{\backslash text\{eye\}\}\; -\; y\_\{\backslash text\{cheek\}\}$

4. Lip Depression Calculation (Sadness)

The lip depression is calculated as the vertical distance between the lip corner and the bottom lip:

$\text{lipDepression}=y_{\text{bottomLip}}-y_{\text{lipCorner}}\backslash text\{lipDepression\}\; =\; y\_\{\backslash text\{bottomLip\}\}\; -\; y\_\{\backslash text\{lipCorner\}\}$

5. Brow Lowering Calculation (Anger)

The brow lowering is calculated as the vertical distance between the inner and outer brow landmarks:

$\text{browLower}=y_{\text{innerBrow}}-y_{\text{outerBrow}}\backslash text\{browLower\}\; =\; y\_\{\backslash text\{innerBrow\}\}\; -\; y\_\{\backslash text\{outerBrow\}\}$

6. Eye Openness Calculation (Surprise)

The eye openness is calculated as the vertical distance between the eyelid and eye landmarks:

$\text{eyeOpenness}=y_{\text{eye}}-y_{\text{eyelid}}\backslash text\{eyeOpenness\}\; =\; y\_\{\backslash text\{eye\}\}\; -\; y\_\{\backslash text\{eyelid\}\}$

7. Jaw Drop Calculation (Surprise)

The jaw drop is calculated as the vertical distance between the chin and nose landmarks:

$\text{jawDrop}=y_{\text{chin}}-y_{\text{nose}}\backslash text\{jawDrop\}\; =\; y\_\{\backslash text\{chin\}\}\; -\; y\_\{\backslash text\{nose\}\}$

8. Deviation from Neutral (Neutral Emotion)

The deviation from neutral is calculated as the sum of Euclidean distances between key facial landmarks:

$\text{deviation}=\sqrt{(x_2-x_1{)}^2+(y_2-y_1{)}^2}\backslash text\{deviation\}\; =\; \backslash sqrt\{(x\_2\; -\; x\_1)^2\; +\; (y\_2\; -\; y\_1)^2\}$

The total deviation from neutral is calculated as the sum of Euclidean distances between key facial landmarks:

$\text{deviation}=\sum _{i=1}^n\sqrt{(x_{2i}-x_{1i}{)}^2+(y_{2i}-y_{1i}{)}^2}\backslash text\{deviation\}\; =\; \backslash sum\_\{i=1\}^\{n\}\; \backslash sqrt\{(x\_\{2i\}\; -\; x\_\{1i\})^2\; +\; (y\_\{2i\}\; -\; y\_\{1i\})^2\}$

Technologies:

• Face Recognition
• Emotion Detection
• Voice Analysis
• IoT with Health Sensor (e.g., GSR Sensor, MAX30102 Sensor, BH1750 Sensor, and ESP32 Microcontroller)
• Machine Learning

Usage:

1. Open the application and allow camera and microphone access.
2. Let the application analyze your facial expressions.
3. Receive tailored recommendations based on your condition.

How To Use this Package

You can copy the command line below:

npm install -g @galihridhoutomo/mentalhealth

Import Modules into Project

If using CommonJS:

const EmotionDetection = require('@galihridhoutomo/mentalhealth');

or If using ES Module (ESM):

import EmotionDetection from '@galihridhoutomo/mentalhealth';

Detecting Emotion from Face Images

Use the detectEmotion(imagePath) function to detect emotions from facial images:

EmotionDetection.detectEmotion('path/to/image.jpg')
.then(result => {
  console.log('Emotion Detection Result:', result);
})
.catch(error => {
  console.error('Error:', error);
});

Sample Output:

{
  "emotion": "happy",
  "confidence": 0.92
}

Detecting Emotion from Camera in Real-Time

If you want to detect emotions from the camera, use the detectEmotionLive() function:

EmotionDetection.detectEmotionLive()
.then(result => {
  console.log('Detected Emotion:', result);
})
.catch(error => {
  console.error('Error:', error);
});

Additional Configuration (Optional)

You can customize the detection model with the following options:

const options = {
  model: 'advanced',  # Can be 'basic' or 'advanced'
  threshold: 0.8      # Minimum confidence threshold
};

EmotionDetection.detectEmotion('path/to/image.jpg', options)
  .then(result => console.log(result))
  .catch(error => console.error(error));

Cite

If you are using this GitHub Repository, please cite it in the following format:

@misc{mentalhealth-app,
  author = {Utomo, Galih Ridho and Maulida, Ana},
  title = {Mental Health Application with Face Recognition and Emotion Detection},
  year = {2025},
  howpublished = {\url{https://github.com/4211421036/MentalHealth}},
  note = {GitHub repository},
}

Authors

1. Galih Ridho Utomo
2. Ana Maulida

---