Human-centric
AI Computer Vision
3DiVi Inc., founded in 2011, is one of the leading developers of AI and machine learning (ML) technologies for computer vision.
Your request has been successfully sent. We'll get in touch shortly.
THANK YOU!

Zero Drunk-Driving with Face Recognition in a Kotlin App for Alcohol Testing

Introduction

One in four fatal road incidents in the EU involves alcohol. Globally, the numbers are just as shocking: out of the 1.25 million annual road deaths, 273,000 involve a drunk driver.

Yet, mandatory alcohol testing remains limited to only a small group of drivers — typically public and commercial transport. Tests are usually done at the start of a shift, under the “watchful” eye of a supervisor.

But what happens when there’s no supervisor, no control, and every test relies purely on trust?

  • What if a driver is hundreds of miles into a solo trip and needs to self-test?

  • What if someone cheats the test — or gets help to do it?

These were exactly the challenges brought to us by a leading breathalyzer vendor in Eastern Europe. They needed a reliable way to verify the identity of the person taking the test — to eliminate substitution and ensure the test is taken by the right person, every time.

Our solution? Face recognition with liveness detection, integrated into the alcohol testing process, making sure every breath test is real, verified, and impossible to fake — no matter where the driver is.

Challenges

Synchronizing Face Recognition with the Test Moment
To eliminate identity fraud, facial verification must occur precisely when the test is being taken — that is, while the user is actively blowing into the breathalyzer. Timing and accuracy are critical here to ensure the right person is tested.

Ensuring Reliability in Harsh Environments
The system needs to perform flawlessly in challenging conditions — from low-light to humidity and dust. Additionally, it must still recognize faces even if partially obstructed by things like glasses or hoods.

Supporting Diverse Workflows
The solution had to cater to two distinct workflows:

  • 1:N Identification for terminals used by multiple employees.

  • 1:1 Verification for mobile apps linked to a specific driver.

Solution

Since the customer’s devices run on Android, integration of 3DiVi Face SDK using the Kotlin API was fast and straightforward. The team implemented two solution formats:

1. Terminal + Breathalyzer

The terminal is a standard Android device with a connected breathalyzer. Once the test begins, the terminal captures a face image and verifies the person in real time. Since the test hardware is physically linked to the terminal, it’s easy to track the exact moment the test starts and trigger recognition.

2. Mobile App + Bluetooth Breathalyzer

For the mobile app, the breathalyzer connects via Bluetooth. The process is similar: face recognition is triggered when the test starts. Since the app is used by a single driver, 1:1 verification is applied.

Implementation

Both the terminal and mobile app use the standard face recognition pipeline:  on the terminal, 1:N identification is employed to identify the correct person from a large database, while the mobile app uses 1:1 verification, as it is designed for use by a single driver.

The terminal version also uses the DynamicTemplateIndex module from 3DiVi Face SDK 3.25.0, which enables flexible management of a dynamic employee database.

Results

The solution was successfully rolled out across three transport companies. Before integration, each company experienced 2–3 incidents per month involving attempts to cheat the alcohol test or drive under the influence. Plus, each company reported 4–5 alcohol-related car accidents each year.

After implementing 3DiVi Face SDK, the transformation was striking:

  • Substitutions became impossible

  • No drunk-driving cases in the first six months

  • Zero alcohol-related accidents in the same period

What’s more, the deployment was lightning fast — taking just 1.5 months from development to production, all managed by a single in-house developer.

An extra bonus? The flexible licensing model allowed the client to use a single ESL license across all Android devices, cutting out per-device costs and making it easy to scale without additional expenses.

Want to add face biometrics to your Kotlin app? Schedule a free consultation to uncover the best integration options for your business.

Read More Success Stories

Omnigo Software
Ecortex
Mobile-Technologies Inc.
Papillon APFIS
Unique Technologies
Papillon ACS
Start your project with 3DiVi