Face Recognition on Android Without the Cloud: A Look at MiniAiLive's On-Premise SDK

You've probably noticed that most face recognition SDKs these days want you to ship your users' biometric data off to a cloud server. That's a hard sell if you're building something for a bank, a government agency, or even just a security-conscious enterprise client. You need the recognition power, but you don't want the data leaving your infrastructure. That's exactly the gap this Android SDK from MiniAiLive is trying to fill.

The project is a fully on-premise face recognition SDK with 3D passive liveness detection (a fancy way of saying it can tell if a face is real or a photo/video mask). All processing happens on your own server, and no data ever leaves it. If you're building an Android app that needs to verify identity—think access control, fintech onboarding, or smart retail—this is worth a closer look.

What It Does

At its core, this is an Android SDK that gives your app two capabilities: face recognition and face liveness detection. The liveness part is the "anti-spoofing" feature—it's designed to detect when someone tries to trick the system with a printed photo, a screen replay, or a 3D mask.

The architecture is straightforward. You get a libfacesdk folder that you drop into your Android project, add it as a module in your settings.gradle file, and you're ready to integrate. The SDK is written in Kotlin (the sample code shows a MainActivity.kt), and it's designed to be called directly from your app's code.

The "on-premise" part is the key differentiator here. The README states it plainly: "SDK is fully on-premise, processing all happens on hosting server and no data leaves server." That means you're not sending biometric data to a third-party API. You're running the recognition logic on hardware you control, which is a huge deal for compliance and privacy.

You'll need a license to use it, but they offer a trial license if you contact them. The README also points to a demo APK on the Google Play Store, so you can see it working before you commit to integrating it.

Why It's Cool

Here's what makes this project stand out from the sea of face recognition libraries:

• Privacy by architecture. The on-premise model isn't a bolt-on feature—it's the whole point. For developers working in regulated industries (fintech, healthcare, government), this is often a non-negotiable requirement. You get the functionality without the data governance headache.

• Passive liveness detection. This is smarter than it sounds. "Passive" means the user doesn't have to do anything weird like blink on command or turn their head in a specific pattern. The SDK just analyzes the face and determines if it's a real person. That's a much better user experience than the active liveness checks you see in some products.

• It's an Android SDK, not a cloud API. You're not dealing with network latency or rate limits. The processing happens on your server, so your app can do face matching and liveness checks as part of its normal flow, without waiting on a round-trip to some external service.

• The demo is live. You can literally download the APK from Google Play and try it before you write a single line of code. That's rare for an SDK and a huge time-saver if you're evaluating it for a project.

• They're responsive. The README mentions they're available 24/7 on WhatsApp for licensing questions. That's a nice touch for a developer tool, especially when you're trying to get a trial license sorted out.

How to Try It

The first thing you'll want to do is grab the SDK. The README links to a Google Drive download for the latest SDK version, and there's also a demo APK on the Play Store if you want to see it in action first.

Once you've downloaded the SDK, the setup is a two-step process:

1. Copy the SDK into your project. Take the libfacesdk folder and drop it into the root folder of your Android project.

2. Add it to your Gradle settings. In your settings.gradle file, add this line:

include ':libfacesdk'

After that, you'll need to request a trial license from their contact page. The README provides a sample license key in the MainActivity.kt file, but you're expected to replace it with your own:

var ret = FaceSDK.setActivation(
    "dYSREvlnlNxuMwFlDCngsmkG5rFIck95ymNvkPDeTUXt3Cj7y0sFIoYIuv3rXaeCb6Imf7lbr7r09S" + 
    "sAPPPhL6oD1uCsdRqddQcMjzHThgjLBXjphSMnclb8SM8mzs/brmMZ+Ofu2p7nqKIy7zJASB3iRo2L5gy7e" + 
    "hNvKUy4bSyt7n7xCz8PrGWmBnphupYbQJLGU24RdVN1suybukqjZX5ctUUu2sDSd2CawEDH7ftLyoLuFrG1v" + 
    "YqExNq/FOhgBjzHgSmZ1tiZa+35rrU6kyzUG6O9Nl8A+Wr/lsV2QDjqn7iGgPmzimGL2pr7OLXJRUkOOWldW" + 
    "detUeqohaI9eA=="
)

The repository also links to their documentation site, a HuggingFace page, and a live demo, so you've got plenty of resources to explore before you dive into the code.

Final Thoughts

This SDK is squarely aimed at developers who need face recognition in their Android apps but can't compromise on data privacy. It's not trying to be a general-purpose computer vision library—it's a focused tool for a specific problem, and it does that job well.

The on-premise requirement means you'll need to handle your own server infrastructure, so this isn't a zero-maintenance drop-in. But if you're building for industries where biometric data can't leave your control, that's a feature, not a bug. The trial license and the Play Store demo make it easy to evaluate, and the responsive support team is a nice bonus.

If you're in the market for face recognition with liveness detection and you've been frustrated by cloud-only options, this is worth a look. You can find the full repository and setup instructions on GitHub.

Follow @githubprojects for more developer tools and open source projects.