🚌 Busbom 버스봄 - Main Server

🏗️ Project Overview

Busbom is a smart bus stop guidance system. It automatically monitors and guides bus arrivals and platform status through real-time video processing, AI-based license plate recognition, and ONVIF metadata analysis.

The system consists of a supporting server platform_observer, hardware driver hardware, and client application Qt_Client.

✨ Key Features

• 🎥 Real-time Video Streaming: Support for RPi cameras and ONVIF IP cameras
• 🔍 AI-based License Plate Recognition: YOLOv5n + TensorFlow Lite OCR
• 📡 RTSP Streaming: Real-time stream transmission based on MediaMTX
• 🚏 Platform Status Management: Real-time monitoring of bus arrivals/departures
• 🌐 API Interface: HTTPS API based on FastCGI + Nginx
• 💾 Shared Memory: High-performance data exchange between processes

📁 Project Structure

rpi_server/
├── cgi/                        # FastCGI Web Applications
│   ├── capture.cgi/            # Platform Video Capture API
│   ├── config.cgi/             # Camera Settings Control API
│   ├── roi-setup.cgi/          # ROI Area Setup API
│   ├── stop-status.cgi/        # Bus Stop Status Query API
│   ├── sequence.cgi/           # Bus Approach Sequence API
│   ├── bus-mapping.cgi/        # Bus Route Mapping API
│   └── sleep.cgi/              # STM32 Power Control API
├── camera_stream/              # RPi Camera Streaming
│   ├── main.cpp                # Frame Shared Memory Storage
│   └── CMakeLists.txt
├── yolo_lp_detector/           # YOLO License Plate Detector (Pi Cam)
│   ├── main.cpp                # YOLO + OCR Main Application
│   ├── yolo.cpp/.hpp           # YOLOv5n NCNN Inference Engine
│   ├── tf_ocr.cpp/.hpp         # TensorFlow Lite OCR Engine
│   ├── plate.cpp/.hpp          # License Plate Preprocessing Utilities
│   └── lp_detect_v5n.ncnn.*    # NCNN Model Files
├── onvif_streamer/             # ONVIF Stream Receiver (Hanwha Camera)
│   ├── main.cpp                # RTSP/Metadata Processing
│   ├── video.cpp/.hpp          # Video Stream Processing
│   ├── parser.cpp/.hpp         # ONVIF Metadata Parsing
│   ├── ocr.cpp/.hpp            # License Plate OCR Processing
│   └── model.tflite            # TensorFlow Lite Model
├── bus_butt/                   # Bus Stop Main Algorithm
│   ├── main.cpp                # Main Algorithm
│   ├── bus_station_manager/    # Bus Stop Status Manager
│   ├── bus_queue_fetcher/      # Bus Approach Queue Query
│   ├── display_writer/         # Hardware Output Management
│   └── stop_status_fetcher/    # Bus Stop Status Collector
├── rtsp_server/                # RTSP Stream Server
│   └── main.cpp                # Shared Memory → RTSP Stream Conversion
└── rtsp_simple_stream/         # Generic RTSP Stream Processor
    ├── main.cpp                # RTSP → Shared Memory Conversion
    └── CMakeLists.txt

🔧 System Requirements

Operating Environment

• Raspberry Pi 4 (8GB RAM recommended)
• RPi Camera Module v2/v3 (optional)
• ONVIF compatible IP camera (optional)

Dependencies

Library	Version	Purpose
ncnn	latest	YOLO inference engine
tensorflow-lite	2.6.x	OCR inference engine
opencv4	4.x	Image processing
mediamtx	1.13+	RTSP server
ffmpeg	4.x+	Video encoding/decoding
tinyxml2	latest	XML metadata parsing
libsdl2	2.x	Real-time video rendering
libfcgi-dev	latest	FastCGI web applications
nginx	1.18+	Web server

🚀 Installation and Execution

1. Clone Repository

git clone https://github.com/BusBom/rpi_server.git
cd rpi_server

2. Install Dependencies

# Update system packages
sudo apt update && sudo apt upgrade -y

# Install essential packages
sudo apt install -y cmake build-essential pkg-config \
    libopencv-dev libsdl2-dev libtinyxml2-dev \
    libfcgi-dev nginx git cmake 

# Install MediaMTX
wget https://github.com/bluenviron/mediamtx/releases/latest/download/mediamtx_v1.13.0_linux_arm64.tar.gz
tar -xzf mediamtx_v1.13.0_linux_arm64.tar.gz
sudo mv mediamtx /usr/local/bin/

3. Build

# Build each module
cd .../build
cmake ..
make

# Build CGI applications
cd cgi
make
sudo make install

4. Configuration

# Nginx configuration
# Separate HTTPS and CGI configuration (conf) setup required
sudo cp nginx.conf /etc/nginx/sites-available/busbom
sudo ln -s /etc/nginx/sites-available/busbom /etc/nginx/sites-enabled/
sudo systemctl restart nginx

# MediaMTX configuration
sudo cp mediamtx.yml /etc/mediamtx/
sudo systemctl enable --now mediamtx

5. Execution

# Run each module (in separate terminals)
cd camera_stream/build && ./camera_stream
cd yolo_lp_detector/build && ./yolo_detector  
cd onvif_streamer/build && ./onvif_streamer

📊 Access and Management

HTTPS Access

https://[RaspberryPi_IP]/

API Endpoints

Endpoint	Method	Description
/cgi-bin/stop-status.cgi	GET	Query bus stop platform status
/cgi-bin/capture.cgi	GET	Capture current video frame
/cgi-bin/config.cgi	POST	Change camera settings
/cgi-bin/roi-setup.cgi	POST	Set ROI area
/cgi-bin/sequence.cgi	GET	Query bus route sequence

RTSP Stream Access

rtsp://[RaspberryPi_IP]:8554/stream