Sixfab | DeepX AI Vision Platform Demo

This interactive demo provides a "DeepX AI Vision Platform" UI with real-time hardware monitoring, multi-stream video processing, and a diverse set of AI inference examples.

Getting Started

Quick Install (One-Liner)

Clone and build the project with a single command. The installer checks for git, installs it if missing, clones this repo, and runs build.sh.

curl -fsSL https://raw.githubusercontent.com/sixfab/deepx-demo-application/master/install.sh | bash

Manual Build

You can use the provided build.sh script to install dependencies, download required AI model resources (resources.zip), and compile the project automatically.

./build.sh

2. Run the Application

Once the build finishes successfully, the application executable will be located in the build/ directory.

./build/cpp_qt_app

Video Source Configuration

The application supports multiple camera and video input types. Video sources are configured in the JSON config files located in src/configs/. Each config file has an input.sources array where you define one or more video sources.

Source Types

USB Camera (camera)

Use a standard USB camera (V4L2 device) as input.

"input": {
    "format": "RGB",
    "sources": [
        {
            "type": "camera",
            "path": "/dev/video0"
        }
    ]
}

To find available USB cameras on your system:

ls /dev/video*

Raspberry Pi Camera (libcamera)

Use a Raspberry Pi Camera Module (CSI) via libcamera. The path format is camera_id:width:height:fps.

"sources": [
    {
        "type": "libcamera",
        "path": "0:1280:720:30",
        "flip": true
    }
]

Path Segment	Description	Example
0	Camera ID	0 (first CSI camera)
1280	Width in pixels	1280
720	Height in pixels	720
30	Frame rate (fps)	30

Set "flip": true if your camera image is upside-down (common with certain mounting orientations).

RTSP Stream (rtsp)

Connect to an IP camera or network video stream via RTSP.

"sources": [
    {
        "type": "rtsp",
        "path": "rtsp://192.168.1.100:554/live"
    }
]

Common RTSP URL patterns by manufacturer:

Manufacturer	Example URL
Generic	rtsp://<ip>:<port>/live
Hikvision	rtsp://<user>:<pass>@<ip>:554/Streaming/Channels/1
Dahua	rtsp://<user>:<pass>@<ip>:554/cam/realmonitor?channel=1&subtype=0
Amcrest	rtsp://<user>:<pass>@<ip>:554/cam/realmonitor?channel=1&subtype=0

Video File (video)

Use a local video file (MP4, AVI, etc.) as input.

"sources": [
    {
        "type": "video",
        "path": "src/videos/traffic.mp4"
    }
]

Image File (image)

Use static image files as input (useful for OCR or single-frame analysis).

"sources": [
    {
        "type": "image",
        "path": "src/images/sample.png"
    }
]

Example Configurations

Config File	Source Type	Description
yolov8n_ppu_libcamera_rpi.json	RPi Camera	Object detection with Raspberry Pi Camera Module
yolov11_ppu_realtime_example.json	USB Camera + RTSP fallback	PPE compliance detection
yolov5s_ppu_mask_detection_realtime.json	Dual USB Camera	Mask detection with cameras
smart_traffic_config.json	Video File + RTSP	Traffic flow monitoring
yolov8n_ppu_realtime_example_trespassing.json	Multi-channel Video	4-channel perimeter intrusion detection
multi_videofiles_36ch.json	Multi-channel Video	36-channel large-scale video analytics
ocr_v4.json	Image + RPi Camera	OCR with static images and live camera