aoguai/LiYing

What it solves

LiYing automates the tedious post-processing workflow of professional photo studios for creating ID photos. It eliminates the need for manual editing by automating face detection, alignment, background removal, and layout arrangement.

How it works

The software uses a pipeline of specialized AI models to process single-person portrait photos:

• Face and Body Detection: Uses YuNet for face recognition and YOLOv8-pose for body identification to ensure correct positioning.
• Angle Correction: Automatically adjusts the angle of the subject.
• Background Processing: Employs RMBG (1.4/2.0) to remove the original background and replace it with any specified color or transparency.
• Cropping and Layout: Automatically crops the image to standard ID photo sizes and arranges multiple copies onto a single photo sheet for printing.
• Inference: Runs locally and offline, supporting both CPU and NVIDIA GPU acceleration via ONNX Runtime.

Who it’s for

• Photo studio operators looking to automate their ID photo workflow.
• Individuals who need to create professional ID photos from home using portrait shots.

Highlights

• Fully Offline: All image processing happens locally for privacy.
• Comprehensive Automation: Handles everything from detection and correction to background replacement and final layout.
• Flexible Output: Supports custom photo sizes, background colors, and specific print sheet layouts (rows/cols).
• Multiple Interfaces: Available via Command Line Interface (CLI), a Gradio-based WebUI, and a pre-packaged Windows bundle.
• Precision Control: Includes options for target file size (KB) and the addition of crop lines on the final sheet.