🏢 Smart Building Floor Access Control System

A multi-layered physical security system for a 4-floor building, combining RFID card access, fingerprint biometrics, and live face recognition, all controlled through a Raspberry Pi GUI dashboard.

---

📸 System Overview

The system enforces a strict entry/exit sequence:

ENTER:  Fingerprint scan (building gate) → RFID tap Floor 1 → RFID tap target floor
EXIT:   RFID tap target floor (OUT) → RFID tap Floor 1 (return to lobby) → Fingerprint scan (gate)

Skipping any step triggers a CRITICAL anomaly alert and denies access.

---

⚙️ Hardware Architecture

Component	Role
Raspberry Pi	Central controller, runs the GUI dashboard
ESP32-S3	Manages all 4 RFID readers + fingerprint sensor, communicates with Pi via USB serial
Arduino Nano	Controls relays (floor lights) and servos (door locks) via single-byte commands from ESP32
MFRC522 x 4	RFID readers, one per floor (shared SPI bus, individual SS pins)
R307S	Fingerprint sensor (UART, up to 127 templates)
PCA9685	I2C servo driver, controls door locks on Floors 2-4
8-channel relay module	White + red lights per floor (Active LOW)
Pi Camera x 2	Live surveillance on Floors 2 and 3

Wiring Quick Reference

ESP32-S3 RFID (shared SPI)

SCK  → GPIO 36 | MISO → GPIO 37 | MOSI → GPIO 35 | RST → GPIO 8
SS1 (F1) → GPIO 4 | SS2 (F2) → GPIO 5 | SS3 (F3) → GPIO 6 | SS4 (F4) → GPIO 7

R307S Fingerprint Sensor

Red → 5V | Black → GND
Yellow (TX) → GPIO 16 | Green (RX) → GPIO 19 | White (Wake) → GPIO 20

ESP32 → Arduino Nano UART

ESP32 TX1 (GPIO 17) → Nano D0 (via 1kΩ resistor)
ESP32 RX1 (GPIO 18) ← Nano D1 | GND ↔ GND

---

🖥️ Software Stack

• Python 3 - Main application (syste1.py)
• CustomTkinter - Dark-themed GUI dashboard
• OpenCV + face_recognition - Real-time face detection and recognition
• Picamera2 - Pi camera integration
• SQLite - Persistent access log, anomaly log, and card state
• PySerial - USB serial communication with ESP32
• Arduino (C++) - Firmware for ESP32-S3 and Arduino Nano

---

🛡️ Security Features

10 Anomaly Types (4 Severity Levels)

Severity	Anomaly	Action Required
🔴 CRITICAL	Unknown Card	Confiscate card, no identity on record
🔴 CRITICAL	No FP Sign-In	Intercept, possible tailgate or stolen card
🔴 CRITICAL	FP Exit Order Violation	Intercept, unaccounted exit
🟠 HIGH	Unauthorized Floor	Escort out, verify clearance
🟠 HIGH	Floor 1 Bypass	Check entry point, possible back-door
🟡 MEDIUM	Out-of-Order Exit	Remind of correct exit sequence
🟡 MEDIUM	Orphan Out	Review entry log
🟡 MEDIUM	Multi-Floor In	Verify occupancy
🔵 LOW	Floor Skip	Verify floor compliance
🔵 LOW	Rapid Re-entry	Check for card sharing

Additional Security Layers

• Strict Mode - denies access on any anomaly detection
• Camera surveillance - cross-references face recognition with RFID state
• Multi-floor face detection - alerts when the same person appears on two camera feeds simultaneously
• Stale-state sweep - hourly check flags anyone marked "inside" for 24+ hours
• Emergency/Evacuation Mode - grants all access unconditionally, loops alarm sound

---

📋 Dashboard Pages

Page	Description
Dashboard	Live floor occupancy tiles, light/sensor controls, FP sensor status
Live Feed	Real-time timestamped event log
Camera	Dual camera surveillance with face recognition overlay
Personnel	Employee management, RFID + fingerprint enrollment + face data capture
Access Log	Full access history with granted/denied status
Building Activity	Per-employee location, state, and violation history, undo violations without FP gate
Anomaly Log	Security violations with severity, action guidance, and strict mode controls
Visitors	External visitor request approval workflow
Temp Access	Employee temporary floor access requests with countdown timer
System	Connection, time sync, emergency mode, diagnostics

---

🚀 Setup & Installation

1. Python Dependencies

pip install customtkinter pyserial opencv-python pillow face_recognition picamera2 numpy

face_recognition requires dlib. On Raspberry Pi:

pip install cmake dlib face_recognition

2. Arduino Firmware

• ESP32-S3 - Flash floor_access_fingerprint_v4.ino using Arduino IDE

  • Board: ESP32S3 Dev Module
  • Flash: 16MB, PSRAM: OPI PSRAM (8MB)
  • Required library: Adafruit Fingerprint Sensor Library

• Arduino Nano - Flash nano_relay_servo_controller_fixed__1__ino.ino

  • Required library: Adafruit PWMServoDriver (for PCA9685)

3. Face Recognition Setup

Create a dataset/ folder with one sub-folder per employee (named exactly as registered in the system):

dataset/
  Juan Dela Cruz/
    photo1.jpg
    photo2.jpg
  Maria Santos/
    photo1.jpg

Then train the model from the Personnel page (🧠 TRAIN MODEL button) or via the Face Capture dialog.

4. Audio Files

Place WAV files in an audios/ folder next to syste1.py:

audios/
  grant.wav
  grant_fp.wav
  grant_floor.wav
  grant_floor1.wav
  grant_temp.wav
  deny.wav
  deny_unknown_card.wav
  deny_fp_required.wav
  deny_sequence.wav
  violation.wav
  alert_fp_missing.wav
  alert_intercept.wav
  alert_bypass.wav
  alert_out_of_order.wav
  alert_multi_floor.wav
  alert_floor_skip.wav
  emergency.wav

5. Run

python3 syste1.py

The app auto-scans for the ESP32 on startup. Use the PORT dropdown to select manually if needed.

---

📡 ESP32 ↔ Pi Protocol

Pi → ESP32 (text commands):

ADD:<uid>,<floor>,<name>     # Register RFID card
DEL:<uid>                    # Remove card
LIST_CARDS                   # Fetch all cards
FP_ENROLL:<id>,<floor>,<name>  # Start fingerprint enrollment
FP_DELETE:<id>               # Delete fingerprint
GRANT_FLOOR:<floor>          # Unlock door for one access
LIGHT:<floor>,<normal|alert> # Override floor lighting
EMERGENCY:<ON|OFF>           # Toggle evacuation mode
TIME:<epoch>                 # Sync RTC

ESP32 → Pi (JSON events):

{"event": "scan",    "floor": 2, "uid": "AA BB CC DD", "result": "GRANTED", "name": "Juan", "dir": "IN",  "time": "2025-01-01 08:00:00"}
{"event": "fp_scan", "floor": 0, "fp_id": 1, "confidence": 250, "result": "GRANTED", "name": "Juan", "dir": "IN"}
{"event": "fp_status", "status": "ready", "count": 3}

---

🗂️ File Structure

smart building security/
├── syste1.py                          # Main Raspberry Pi application
├── floor_access_fingerprint_v4.ino    # ESP32-S3 firmware
├── nano_relay_servo_controller_fixed__1__ino/
│   └── nano_relay_servo_controller_fixed__1__ino.ino  # Arduino Nano firmware
├── audios/                            # WAV audio files (not tracked)
├── dataset/                           # Face recognition training images (not tracked)
├── captured_faces/                    # Auto-captured face snapshots (not tracked)
├── encodings.pickle                   # Trained face model (not tracked, auto-generated)
└── floor_access.db                    # SQLite database (not tracked, auto-generated)

---

.gitignore Recommendations

# Auto-generated
floor_access.db
encodings.pickle
security_system.log

# Face data (privacy)
dataset/
captured_faces/

# Audio assets
audios/

---

📝 Notes

• Baud rate between Pi and ESP32: 115200
• Maximum registered employees: 50 (RFID) / 127 (fingerprint templates)
• Supports up to 4 floors
• Face recognition uses HOG model (face_recognition library) scaled at 1/3 resolution for performance
• All camera UI updates are marshalled to the main Tkinter thread via after(0, ...) for thread safety
• Database uses threading.RLock for re-entrant access between the main event loop and camera thread