🔴 Security Camera Weapon Detection System

A real-time AI-powered security camera system that detects weapons (pistols, knives, rifles, grenades) using your webcam or IP camera. When a weapon is detected, it automatically:

📸 Takes a snapshot of the person
🎥 Records a video clip
📧 Sends an email alert with photo
📱 Sends an SMS (via Twilio)
🤖 Sends a Telegram alert with photo
🔊 Plays an alarm sound
🌐 Shows everything on a live web dashboard

📸 Real Detection Screenshots

These are actual detections captured by the system:

Pistol Detection	Knife Detection

🧠 How It Works (Simple Explanation)

Your Webcam
    │
    ▼
┌─────────────────────────────────────────────┐
│  Frame captured every ~33ms                 │
│                                             │
│  1. Night Mode (auto-brighten if dark)      │
│  2. Face Detection  (built-in OpenCV)       │
│  3. Person Detection  (YOLOv8 — local)      │
│  4. Weapon Detection  (Roboflow AI — cloud) │
└─────────────────────────────────────────────┘
    │
    ▼  Weapon found with confidence ≥ 35%
    │  AND detected in 3+ consecutive frames?
    │
    ▼
┌─────────────────────────────────────────────┐
│  Alert Pipeline (all run in background)     │
│                                             │
│  📸 Save person snapshot                   │
│  🎥 Record 10-second video clip            │
│  🔊 Play alarm beep                        │
│  📧 Send email with photo                  │
│  📱 Send SMS                               │
│  🤖 Send Telegram message + photo          │
│  💾 Log to SQLite database                 │
└─────────────────────────────────────────────┘
    │
    ▼
🌐 Web Dashboard at http://localhost:5000
   (live camera feed + alert history table)

📁 Project Structure

Security_cam/
│
├── main.py              ← Run this to start the system
├── config.py            ← ALL your settings go here
│
├── detector.py          ← Sends frames to Roboflow AI for weapon detection
├── dashboard.py         ← Flask web dashboard (live feed + alert log)
├── database.py          ← SQLite database (stores all alerts)
├── emailer.py           ← Sends email alerts with photo attached
├── sms_alert.py         ← Sends SMS via Twilio
├── telegram_alert.py    ← Sends Telegram alerts with photo
├── snapshot.py          ← Saves person photo + records video clip
├── alarm.py             ← Plays beep alarm sound
├── scheduler.py         ← Limits active hours + sends daily report
│
├── yolov8n.pt           ← YOLOv8 model (person detection, runs locally)
├── security_log.db      ← Auto-created SQLite database
├── snapshots/           ← Auto-created folder for saved photos
└── videos/              ← Auto-created folder for saved video clips

✅ Requirements

Python 3.8 or higher
A webcam (built-in laptop cam works fine)
A free Roboflow account (for weapon detection AI)
Optional: Gmail account, Twilio account, Telegram bot

🚀 Quick Start (Step by Step)

Step 1 — Clone the Repository

git clone https://github.com/YOUR_USERNAME/Security_cam.git
cd Security_cam

Step 2 — Install Python Dependencies

pip install opencv-python ultralytics flask requests numpy

If you want SMS alerts, also install:

pip install twilio

Step 3 — Get Your Roboflow API Key (FREE)

This is the AI engine that detects weapons. It's free to use.

Go to https://roboflow.com and create a free account
After logging in, click your profile icon → Settings → API Keys
Copy your Private API Key

Step 4 — Edit `config.py`

Open config.py and fill in your details:

# ── Roboflow (REQUIRED) ────────────────────────
ROBOFLOW_API_KEY  = "paste_your_api_key_here"   # ← Your Roboflow API key
ROBOFLOW_MODEL    = "weapon-detection-f1lih/1"  # ← Keep this as-is

# ── Email (OPTIONAL) ──────────────────────────
EMAIL_SENDER    = "[email protected]"         # ← Your Gmail address
EMAIL_PASSWORD  = "your_app_password"            # ← Gmail App Password (not your real password)
EMAIL_RECEIVERS = ["[email protected]"]        # ← Who gets alerts

# ── Camera ────────────────────────────────────
CAMERA_INDEXES = [0]   # ← 0 = built-in webcam, 1 = second camera

For EMAIL_PASSWORD: Gmail requires an "App Password" (not your normal password). Go to your Google Account → Security → 2-Step Verification → App Passwords → Generate one.

Step 5 — Run the System

python main.py

You will see:

[DB] Initialized.
[INFO] Loading person detection model...
[DETECTOR] Background detection thread started.
[DASHBOARD] Running at http://localhost:5000
[INFO] Starting 1 camera(s)...
[INFO] Dashboard → http://localhost:5000
[INFO] Press Q in any camera window to quit.
[CAM 0] Opened.

A camera window will open showing the live feed with detection boxes
Open your browser at http://localhost:5000 for the web dashboard
Press Q in the camera window to quit

⚙️ All Settings Explained (`config.py`)

Email Settings

EMAIL_SENDER    = "[email protected]"       # Gmail that sends alerts
EMAIL_PASSWORD  = "xxxx xxxx xxxx xxxx" # Gmail App Password
EMAIL_RECEIVERS = ["[email protected]"]  # Can add multiple recipients
SMTP_HOST       = "smtp.gmail.com"      # Leave as-is for Gmail
SMTP_PORT       = 587                   # Leave as-is

SMS Settings (Twilio)

TWILIO_ENABLED = False          # Change to True to enable
TWILIO_SID     = "ACxxx..."     # From your Twilio dashboard
TWILIO_AUTH    = "your_token"   # From your Twilio dashboard
TWILIO_FROM    = "+1xxxxxxxxxx" # Your Twilio phone number
TWILIO_TO      = ["+91xxxxxxx"] # Recipients (any country)

Telegram Settings

TELEGRAM_ENABLED  = False           # Change to True to enable
TELEGRAM_TOKEN    = "123:ABCxxx"    # From @BotFather on Telegram
TELEGRAM_CHAT_IDS = ["123456789"]   # Your chat ID (use @userinfobot)

How to get a Telegram Bot Token:

Open Telegram, search for @BotFather

Send /newbot and follow instructions

Copy the token it gives you

Camera Settings

CAMERA_INDEXES = [0]      # [0] = webcam, [0, 1] = two cameras
FRAME_WIDTH    = 640      # Resolution width
FRAME_HEIGHT   = 480      # Resolution height
AUTO_RECONNECT = True     # Auto-reconnect if camera disconnects
NIGHT_MODE     = False    # Auto-brighten in dark environments

Detection Settings

CONFIDENCE        = 0.35  # Minimum confidence (0.0–1.0). Lower = more sensitive
ALERT_COOLDOWN    = 30    # Seconds to wait before re-alerting for same weapon
MIN_DETECT_FRAMES = 3     # Weapon must appear in 3 consecutive frames (reduces false alarms)
MOTION_DETECTION  = True  # Only call AI when motion is detected (saves API calls)
MOTION_THRESHOLD  = 5000  # Motion sensitivity — lower number = more sensitive

Video & Alarm

SAVE_VIDEO_CLIP    = True  # Save a video clip when weapon detected
VIDEO_CLIP_SECONDS = 10    # Length of video clip in seconds
ALARM_ENABLED      = True  # Beep sound when weapon detected
ALARM_DURATION     = 3     # How long the alarm beeps (seconds)

Schedule (only run at certain hours)

SCHEDULE_ENABLED    = False  # True = only run during set hours
SCHEDULE_START_HOUR = 22     # 10 PM
SCHEDULE_END_HOUR   = 6      # 6 AM (supports overnight schedules)

Daily Report

DAILY_REPORT_ENABLED = True  # Send a summary email every day
DAILY_REPORT_HOUR    = 8     # Send at 8 AM

Weapon Classes

WEAPON_CLASSES = {"pistol", "knife", "rifle", "grenade", "missile"}
# These are the labels the AI model can detect

🌐 Web Dashboard

Open http://localhost:5000 in your browser after starting the system.

Features:

📷 Live camera feed(s) with real-time video stream
📋 Alert log table showing all detections
🟢 Live indicator — auto-refreshes every 3 seconds (no page reload needed)
Shows: ID, Time, Camera, Weapon type, Snapshot path, Email/SMS/Telegram status

🗄️ View Saved Alerts (Database)

To print all logged alerts in the terminal:

python database.py

Output example:

ID    Time                   Cam   Weapon       Snapshot
----------------------------------------------------------------------
3     2026-03-01 22:04:58    0     pistol       snapshots/pistol_20260301_220449.jpg
2     2026-03-01 22:04:49    0     knife        snapshots/knife_20260301_220449.jpg
1     2026-03-01 21:12:28    0     pistol       snapshots/pistol_20260301_211228.jpg

Total alerts: 3

🔧 Troubleshooting

Camera not opening

[CAM 0] Opened.   ← Good

If it doesn't open, try CAMERA_INDEXES = [1] in config.py (some laptops use index 1).

No weapon detections

Make sure ROBOFLOW_API_KEY is set correctly in config.py
Check your internet connection (Roboflow runs in the cloud)
Lower the confidence: CONFIDENCE = 0.25
Lower MIN_DETECT_FRAMES: MIN_DETECT_FRAMES = 1 (more sensitive, more false alarms)

Email not sending

Make sure you are using a Gmail App Password, NOT your regular password
Enable 2-Step Verification on your Google account first
Check that EMAIL_SENDER and EMAIL_PASSWORD are correct

`ModuleNotFoundError`

Install missing packages:

pip install opencv-python ultralytics flask requests numpy

Dashboard not loading

Make sure port 5000 is not already in use. You can change the port in dashboard.py:

app.run(host="0.0.0.0", port=5001, ...)  # change to any free port

📦 Full Dependency List

Package	Purpose	Install
`opencv-python`	Camera capture, face detection, drawing boxes	`pip install opencv-python`
`ultralytics`	YOLOv8 person detection (runs locally)	`pip install ultralytics`
`flask`	Web dashboard server	`pip install flask`
`requests`	Roboflow API calls, Telegram API calls	`pip install requests`
`numpy`	Image processing (motion detection)	`pip install numpy`
`twilio`	SMS alerts (optional)	`pip install twilio`

All at once:

pip install opencv-python ultralytics flask requests numpy twilio

🔑 API Keys Summary

Service	Required?	Where to get it	Config key
Roboflow	YES	roboflow.com → Settings → API Keys	`ROBOFLOW_API_KEY`
Gmail App Password	Optional	Google Account → Security → App Passwords	`EMAIL_PASSWORD`
Twilio	Optional	twilio.com → Console	`TWILIO_SID`, `TWILIO_AUTH`
Telegram Bot	Optional	Telegram → @BotFather → /newbot	`TELEGRAM_TOKEN`

🖥️ Multiple Cameras

To use 2 cameras simultaneously:

# config.py
CAMERA_INDEXES = [0, 1]   # runs both cameras in separate threads

Each camera gets its own window and its own feed in the dashboard.

📂 Output Files

Type	Location	Example
Snapshots	`snapshots/`	`snapshots/pistol_20260301_211228.jpg`
Video clips	`videos/`	`videos/knife_20260301_215628.avi`
Database	`security_log.db`	SQLite, view with `python database.py`

⚡ Quick Command Reference

# Start the system
python main.py

# View all logged alerts
python database.py

# Run only the dashboard (to view existing logs)
python dashboard.py

🛡️ Legal & Ethical Notice

This system is intended for legitimate security monitoring on property you own or have permission to monitor. Always comply with local privacy laws and regulations regarding surveillance. Do not use this system to monitor people without their knowledge where prohibited by law.

🤝 Contributing

Pull requests are welcome. For major changes, open an issue first to discuss what you'd like to change.

Built with OpenCV, YOLOv8, Roboflow, Flask, and Python.

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
__pycache__		__pycache__
snapshots		snapshots
videos		videos
README.md		README.md
alarm.py		alarm.py
config.py		config.py
dashboard.py		dashboard.py
database.py		database.py
detector.py		detector.py
emailer.py		emailer.py
main.py		main.py
scheduler.py		scheduler.py
security_log.db		security_log.db
sms_alert.py		sms_alert.py
snapshot.py		snapshot.py
telegram_alert.py		telegram_alert.py
yolov8n.pt		yolov8n.pt

Folders and files

Latest commit

History

Repository files navigation

🔴 Security Camera Weapon Detection System

📸 Real Detection Screenshots

🧠 How It Works (Simple Explanation)

📁 Project Structure

✅ Requirements

🚀 Quick Start (Step by Step)

Step 1 — Clone the Repository

Step 2 — Install Python Dependencies

Step 3 — Get Your Roboflow API Key (FREE)

Step 4 — Edit config.py

Step 5 — Run the System

⚙️ All Settings Explained (config.py)

Email Settings

SMS Settings (Twilio)

Telegram Settings

Camera Settings

Detection Settings

Video & Alarm

Schedule (only run at certain hours)

Daily Report

Weapon Classes

🌐 Web Dashboard

🗄️ View Saved Alerts (Database)

🔧 Troubleshooting

Camera not opening

No weapon detections

Email not sending

ModuleNotFoundError

Dashboard not loading

📦 Full Dependency List

🔑 API Keys Summary

🖥️ Multiple Cameras

📂 Output Files

⚡ Quick Command Reference

🛡️ Legal & Ethical Notice

🤝 Contributing

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Step 4 — Edit `config.py`

⚙️ All Settings Explained (`config.py`)

`ModuleNotFoundError`

Packages