Skip to content
日昇 Docs

USB Night Vision Camera Module

The USB Night Vision Camera Module is a UVC (USB Video Class) compliant night vision camera capable of capturing clear video day and night. It is recognized driver-free in embedded Linux environments and on PCs, and can be easily integrated into AI image recognition and surveillance systems.

Hardware Specifications

ItemSpecification
SensorHigh-sensitivity CMOS sensor (IR-capable)
Resolution1920x1080 (Full HD)
InterfaceUSB 2.0 / 3.0 (UVC compliant)
IR LEDBuilt-in (nighttime illumination range approx. 10m)
LensWide-angle lens (horizontal FOV approx. 90deg)
PowerUSB bus power (5V)
Operating Temperature-10degC to 50degC
Supported OSLinux, Windows, macOS (UVC driver)

Key Features

Day/Night Vision

With a high-sensitivity CMOS sensor and IR LEDs, the module delivers color video during the day and automatically switches to infrared mode at night, capturing clear monochrome video even in darkness.

UVC Compliant (Driver-Free)

As a USB Video Class compliant device, it is recognized by standard UVC drivers on Linux, Windows, and macOS with no special driver installation required.

Embedded Linux Support

Video can be captured via V4L2 (Video4Linux2), making it easy to build pipelines in combination with GStreamer, FFmpeg, OpenCV, and AI inference frameworks.

Use Cases

  • Nighttime surveillance and security cameras
  • Intrusion detection combined with AI image recognition
  • Nighttime monitoring in nursing care facilities
  • Remote nighttime monitoring of factory equipment
  • Night vision surveillance for parking lots and warehouses
  • PoC for embedded AI camera development

Usage in Linux Environments

Capturing Video with V4L2

Terminal window
# Check device recognition
ls /dev/video*
v4l2-ctl --list-devices


# Check resolution and format
v4l2-ctl -d /dev/video0 --list-formats-ext


# Frame grab
v4l2-ctl -d /dev/video0 --set-fmt-video=width=1920,height=1080,pixelformat=MJPG \
  --stream-mmap --stream-count=1 --stream-to=frame.jpg

GStreamer Pipeline

Terminal window
# Live preview
gst-launch-1.0 v4l2src device=/dev/video0 ! videoconvert ! autovideosink


# H.264 encoding + RTSP streaming
gst-launch-1.0 v4l2src device=/dev/video0 ! videoconvert ! x264enc ! \
  rtph264pay ! udpsink host=192.168.1.100 port=5000

OpenCV Integration

import cv2
cap = cv2.VideoCapture(0)
ret, frame = cap.read()
cv2.imwrite("capture.jpg", frame)
cap.release()

Integration with AI Image Recognition

This module can be used in combination with the following AI recognition processes:

  • Person detection (YOLO / MobileNet SSD)
  • Intrusion detection (zone judgment + object tracking)
  • Vehicle detection (parking lot surveillance)
  • Face detection (entry/exit management)
  • 24-hour operation with night vision + AI recognition

For details, refer to the AI Recognition Algorithms page.

PoC Verification Checklist

Before deployment, it is recommended to verify the following items:

Installation Environment

  • Illuminance at the installation site (day/night)
  • IR illumination range and target area
  • USB cable length (max 5m, extendable with repeaters)
  • Power supply stability
  • Consider housing if waterproof/dustproof requirements apply

Video Quality

  • Day/night video clarity
  • Presence of halation from IR reflections
  • Frame rate (recommended 15fps or higher)
  • Field of view suitability

System Integration

  • Verify V4L2 / UVC recognition
  • Verify video capture with GStreamer / OpenCV
  • Integration verification with AI inference pipeline
  • Integration with recording and streaming functions

Frequently Asked Questions

What should I check if the device is not recognized on Linux?

Use lsusb to check whether the device is enumerated, and check the UVC driver log with dmesg | tail. Trying a different USB cable or port is also effective.

What if the infrared video is overexposed at night?

This can be improved by adjusting IR LED intensity, changing the camera angle, or removing reflective objects. Some modules allow PWM control of LED brightness.

For person detection applications, 5-10fps is often sufficient, but 15fps or higher is recommended for object tracking or fast-moving objects.