LogoVisionLog

Layer 1: Input

Video capture from cameras, webcams, and video files

The Input Layer is the entry point for all video sources into the VisionLog system. It handles video capture from webcams, video files, and image files.

Overview

  • Purpose: Capture video/image input for face processing
  • Input Types: Webcam, Video Files, Images
  • Library: OpenCV

Supported Input Sources

Webcam

Direct-connected cameras for real-time processing.

FeatureSpecification
ConnectionUSB
Resolution1280x720 (default)
Frame Rate30 FPS
Use CaseReal-time attendance, enrollment

Video Files

Offline processing of recorded footage (CCTV, recordings).

FormatExtensionDescription
MP4.mp4Most common, H.264/H.265
AVI.aviWindows standard
MKV.mkvHigh quality container
MOV.movApple format
WMV.wmvWindows Media

Image Files

Single image processing for enrollment or recognition.

FormatExtensions
JPEG.jpg, .jpeg
PNG.png
BMP.bmp
WebP.webp

Input Modes

ModeDescription
CameraReal-time recognition from webcam
VideoProcess recorded video files
ImageSingle image analysis
FolderBatch process folder of images

Frame Skipping

For video processing, frames can be skipped for performance optimization.

Skip ValueUse Case
1Process all frames (maximum accuracy)
5Balanced (default)
10-30Fast scan mode

Best Practices

Camera Placement

  • Height: Face level (1.5-1.8m)
  • Distance: 0.5-3 meters for clear faces
  • Lighting: Even, front-facing light
  • Angle: Direct face view preferred
  • Background: Avoid busy patterns

Image Quality

FactorRecommendation
Minimum Face Size50x50 pixels
Resolution720p or higher
Lighting>300 lux
FocusSharp, not blurry

Video Processing

ScenarioFrame SkipNotes
Real-time1Process all frames
Standard5Balanced (default)
Fast scan10-30Quick overview

Configuration

SettingDescriptionDefault
Camera IndexDefault camera device0
Frame WidthCapture resolution width1280
Frame HeightCapture resolution height720
FPSTarget frame rate30
Frame SkipProcess every Nth frame5

On this page