CCTV Monitor

A full-screen Python GUI application that previews up to 4 RTSP video streams simultaneously in a 2×2 grid.
Designed and optimised for the TSPI RK3566 board (TaishanPi), but runs on any platform that supports Python 3.7+.

---

Features

Feature	Detail
Full-screen 2×2 grid	4 camera panels, letterboxed aspect ratio
Dual-timezone clock	China time (green) + UK time (blue), top-left of title bar
Auto-reconnect	Detects stream loss and reconnects automatically
Hardware decode	CUDA · OpenCL/Mali · MPP (RK3566) · VAAPI · DXVA

---

Screenshots

4-channel grid view	Connection / status overlay

Screenshots taken on a TSPI RK3566 (TaishanPi) running Debian.

---

Hardware Acceleration

The application detects and uses the best available backend at runtime:

Platform	Backend	CPU reduction
NVIDIA GPU	CUDA	60–80 %
RK3566 (TaishanPi)	MPP	70–85 %
Mali GPU	OpenCL	40–60 %
Intel GPU / Linux	VAAPI	40–60 %
Windows	DXVA	40–60 %

RK3566 / MPP notes

• GPU (Mali G52) handles image scaling and rendering via OpenCL (cv2.UMat).
• MPP (Media Process Platform) handles H.264/H.265 video decoding in dedicated hardware.
• NPU is not required — this project does no AI inference.
• MPP is activated automatically when /dev/mpp_service is present or the device tree reports rk3566.

---

Requirements

• Python 3.7+
• A graphical display (X11 / Wayland / Windows desktop)
• Recommended RAM: 1 GB+ (ARM), 2 GB+ (desktop)

---

Installation

Desktop (Windows / Linux / macOS)

git clone https://github.com/nickfu/tspi_CCTV.git
cd tspi_CCTV
pip install -r requirements.txt
python cctv_monitor.py

On Windows you can also double-click run.bat.

ARM device (TaishanPi / Raspberry Pi / Orange Pi)

On your Windows PC:

package_for_arm.bat

This creates cctv_monitor_arm.zip. Transfer it to the ARM device, then:

unzip cctv_monitor_arm.zip
cd cctv_monitor_arm_deploy
sudo bash install_arm.sh

Optional — run on boot:

sudo bash systemd_service.sh
sudo systemctl start cctv-monitor

---

Configuration

Edit cctv_monitor.py and update the video_sources list:

self.video_sources = [
    "rtsp://192.168.1.100/stream1",   # Camera 1
    "rtsp://192.168.1.101/stream1",   # Camera 2
    "rtsp://192.168.1.102/stream1",   # Camera 3
    "rtsp://192.168.1.103/stream1",   # Camera 4
]

Set a source to None to leave that slot empty.

Auto-reconnect settings (also in __init__):

self.reconnect_delay = 3   # seconds between reconnect attempts
self.frame_timeout   = 10  # seconds without a frame = stream lost

---

Keyboard shortcuts

Key	Action
ESC	Exit fullscreen
F11	Toggle fullscreen
Quit button	Close application

---

Troubleshooting

No video / black screen
→ Verify the RTSP URL with VLC or ffplay.
→ Check network connectivity and firewall rules.

High CPU usage
→ Confirm hardware acceleration is active in the startup log (HW Type: mpp).
→ Lower the source resolution on the camera.

MPP not detected on RK3566
→ Check /dev/mpp_service exists (ls -l /dev/mpp_service).
→ Check the kernel module: lsmod | grep mpp.
→ Install the Rockchip MPP library: apt-get install librockchip-mpp1.

---

Project files

cctv_monitor.py        Main application
requirements.txt       Python dependencies (desktop)
requirements_arm.txt   Python dependencies (ARM)
install_arm.sh         ARM installer script
systemd_service.sh     systemd auto-start setup
package_for_arm.bat    Windows helper to create ARM deploy package
run.bat                Windows quick-launch script
setup.py               pip / PyInstaller package config
README.md              This file
demo/pic1.png          Screenshot — 4-channel grid view
demo/pic2.jpg          Screenshot — connection/status overlay

---

Tech stack

• Tkinter — GUI framework
• OpenCV — video capture and hardware-accelerated resize
• Pillow — image conversion for Tkinter display
• pytz — timezone handling
• threading — per-camera background capture threads

---

Licence

MIT