MEMO - Vision-First AI Desktop Companion

MEMO (Memory & Environmental Monitoring Observer) is an intelligent, vision-based desktop companion designed to enhance productivity, health, and spatial awareness. It transforms your standard webcam or mobile phone camera into a smart observer that tracks objects, monitors your posture, guards your focus, and recognizes you.

Built with YOLOv8, FaceNet, and OpenCV, MEMO runs entirely locally on your machine, ensuring privacy and speed.

---

🚀 Features

🧠 Spatial Memory & Querying

MEMO "remembers" where it saw objects on your desk.

• "Where is my bottle?" → "The bottle is currently on the left, seen just now."
• "Do you see my wallet?" → "Yes, I see the wallet."
• Supports fuzzy matching and synonyms (e.g., "phone" = "cell phone").

👤 Face Recognition & Greeting

MEMO knows who you are.

• Personalized Greetings: "Hello Jayadeep. Welcome back."
• Privacy-First: Face embeddings are stored locally (user_embedding.npy) and never uploaded.
• Smart Re-Greeting: Only greets you again if you've been away for a while.

📵 Smart Focus Mode

Need to get work done? Tell MEMO to guard your focus.

• Command: focus on
• Behavior: If MEMO sees a cell phone in the frame, it verbally scolds you: "Put the phone away and focus on your work!"
• Visuals: Bounding boxes turn RED when threats (phones) are detected.

🧘 Health & Posture Coach

MEMO watches out for your physical well-being.

• Proximity Alert: If you lean too close to the screen (>55% frame width), it warns: "You are too close to the screen. Please move back."
• Sedentary Alert: Tracks how long you've been sitting. If > 60 minutes, it reminds you to stretch.

---

🛠️ Technology Stack

• Vision: YOLOv8 (Object Detection & Pose Estimation), FaceNet (Face Recognition).
• Core: OpenCV (Video Processing), NumPy.
• Audio: pyttsx3 (Text-to-Speech).
• Architecture: Modular Python design (Perception → State → Reasoning → Interface).

---

📦 Installation

1. Clone the Repository

   git clone https://github.com/jay7-tech/memo.git
   cd memo

2. Install Dependencies MEMO requires Python 3.8+ and PyTorch.

   pip install -r requirements.txt

   Note: Visual C++ Build Tools may be required for some dependencies on Windows.

🎮 Usage

1. Run the System:

   # Default Webcam (Source 0)
   python main.py
   
   # IP Camera (Example)
   python main.py http://192.168.1.5:8080/video 90

2. Web Dashboard 🌐: Visit http://localhost:5000 to view the live feed and control the system remotely.

3. Voice Commands 🗣️:

   • Toggle Voice: Press v or type voice on.
   • Commands:
     • "Focus on" / "Focus off"
     • "Register me"
     • "Where is my [object]?"
     • "Selfie" (Takes a photo)

4. Keyboard Shortcuts:

   • q: Quit
   • v: Toggle Voice
   • f: Toggle Focus Mode
   • s: Take Selfie

---

🚀 Future Roadmap

See the complete feature roadmap for Raspberry Pi 4B deployment:

📄 docs/FEATURE_ROADMAP.md

This includes:

• 📸 Enhanced Vision: Emotion detection, hand gestures, eye gaze tracking
• 🤖 Robot Behaviors: Head tracking gimbal, idle animations, personality modes
• 🧠 AI Integration: Gemini chat, vision-language models, conversation memory
• 🎛️ Hardware: LED eyes, environment sensors, better audio
• 🌐 Dashboard: Live stats, object map, settings panel

---

Author: [Jayadeep / Jay7-Tech]