Practical overview
When it fits
Use it when building a Windows or Linux Python application that needs image-based face extraction and comparison as SDK components.
Who it is for
- Developers integrating image-based face analysis into Python applications.
- Teams building face comparison or identity-check workflows that need local processing.
Documented uses
- Detect faces and retrieve bounding boxes, landmarks and embeddings from an image.
- Compare face embeddings and obtain a similarity score.
- Add listed face-analysis functions such as liveness, anti-spoofing, pose or expression analysis to an application.
Requirements & limits
The repository identifies the SDK as open source. Its stated platforms are Windows and Linux, and its listed prerequisite is Python 3.9 or later.
- The README describes an on-premise processing model and says GPU use is optional.
- GetImageInfo returns bounding boxes, landmarks and embeddings; get_similarity compares embeddings on a 0–100 scale.
- The README's threshold of 75 is an example setting, not a universal identity-decision standard.
- Anaconda is recommended for dependency management, while the documented installation uses requirements.txt.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up the Python SDK environment
Create the documented Python environment and install the project dependencies.
Before you start
- Windows or Linux
- Python 3.9 or later
- Anaconda is recommended
- A copy of the repository containing requirements.txt
Steps
Create and activate a conda environment with Python 3.9.
Install dependencies with pip install -r requirements.txt.
Run python run.py as the README's installation check.
Expected outcome
The README describes a prepared environment and a command to check the installation.
Guide limitations
- The source lists a repository command for checking installation but does not establish results for a particular machine.
- The guide covers setup and the stated installation check, not deployment in an application.
Extract face data and compare two images
Follow the README example to obtain face embeddings from two images and calculate their similarity.
Before you start
- The SDK installed in a Python 3.9-or-later environment on Windows or Linux
- Two image files; the README lists JPG, PNG, BMP and TIFF formats
Steps
Import FaceRecognition from face_recognition_sdk and initialize it.
Call GetImageInfo for each image with faceMaxCount=1.
If both calls return face information, pass the first embedding from each result to get_similarity.
Optionally compare the returned score with the README's example threshold of 75.
Expected outcome
The documented workflow returns a similarity score from 0 to 100 for the two face embeddings.
Guide limitations
- The threshold of 75 is an example, not a validated universal decision rule.
- The guide follows the README's example and does not establish accuracy or suitability for a specific identity-verification use.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 16 October 2026
- Original website title
- Faceplugin-ltd/Open-Source-Face-Recognition-SDK · Official homepage title · retained in its source language