🧠 MindTrack – Multimodal Emotion Detection System

MindTrack is an AI-powered mental health monitoring system that analyzes both text and facial expressions to detect a user's emotional state.
By combining Natural Language Processing (NLP) and Computer Vision, MindTrack provides real-time insights that help track emotional well-being and identify early signs of mental distress.

---

🌟 Key Features

• 🔤 Text Emotion Analysis
  Detect emotions such as joy, sadness, anger, fear, and more from user-written text.

• 🙂 Image Emotion Analysis
  Identifies facial expressions using a deep learning model trained on emotion datasets.

• 🎯 Multimodal Fusion
  Uses both text and image inputs (optional individually or combined).

• ⚡ Real-Time Prediction
  Fast inference using TensorFlow/Keras models.

• 🗂️ Lightweight & Portable
  Works using .h5 models and .pkl encoders without requiring a heavy backend.

---

🧰 Tech Stack

Component	Technology
Programming Language	Python
Text Model	LSTM / RNN / Deep Learning
Image Model	CNN-based Emotion Classifier
Frameworks	TensorFlow, Keras
NLP Tools	Tokenizer, Label Encoders

---

📁 Project Structure

MindTrack/
│
├── src/
│   ├── app.py                # Main execution script (text/image prediction)
│   ├── text_model.py         # Text emotion detection pipeline
│   ├── image_model.py        # Image-based emotion detection pipeline
│
├── models/
│   ├── text_emotion_model.h5
│   ├── image_emotion_model.h5
│   ├── tokenizer.pkl
│   ├── label_encoder.pkl
│   ├── image_label_encoder.pkl
│
├── requirements.txt          # Python dependencies
├── README.md                 # Project documentation
├── .gitignore                # Ignore cache & environment files

---

🚀 Getting Started

1️⃣ Install Dependencies

pip install -r requirements.txt

2️⃣ Run the Application

python src/app.py

3️⃣ Provide Input

Depending on your version:

• Upload an image → get emotion
• Type text → get emotion
• Combine both → multimodal emotion result

---

🧠 Model Details

Text Emotion Model

• Trained on labeled emotional text dataset
• Tokenizer + sequence padding
• Uses LSTM/RNN layers
• Outputs class probabilities

Image Emotion Model

• CNN trained on facial expression dataset
• Classifies emotions like:
  • Happy
  • Sad
  • Angry
  • Fear
  • Neutral
  • Surprise

---

🎥 Future Enhancements

• Add a full web dashboard
• Add a chatbot-style interface
• Improve facial pre-processing
• Add audio emotion detection
• Deploy on HuggingFace or Streamlit Cloud

---

🧩 Use Cases

• Mental health tracking
• Student emotional monitoring
• Wellness apps
• AI therapy assistants
• Real-time emotion-based recommendations

---

⭐ If You Like This Project

Please star ⭐ the repository — it helps others discover the project.

---