FAQ

This page contains the majority of frequently asked questions about LIQA

Business

Does LIQA require any backend?

No. All calculations are performed on the frontend using the mobile device's capacities. No server connection is required after LIQA initialization, no image or data is sent, and there are no time delays.

How can I try LIQA?

Follow the Try LIQA: access request instructions suitable for your situation

How can I modify the LIQA settings/strictness?

Please visit the Config page to learn about the available settings.

Does LIQA compress images?

LIQA can save final images in different formats (PNG/JPEG). Please refer to the Config page to learn more about the available formats and compressions.

LIQA limits the maximum image resolution to 2000 pixels along each axis in case if the camera can return higher resolution. This size was derived from the optimal image size for further analysis by different AI methods.

Can LIQA detect a beard/glasses?

Currently, this is not possible.

How can I communicate if I have any issues with LIQA?

If you have not been previously contacted by a Haut.AI manager, please, let us know via the contact form on the haut.ai website. Describe the issue and leave an e-mail so we could reach you as soon as possible to help fix the issues.

Technical

Does LIQA work on desktop / tablets?

Currently LIQA does not support such usage. This is a planned feature.

What is the FPS (frames-per-second) rate on modern devices?

The latest iOS and Android devices can easily handle LIQA with 30+ FPS rate. Please see this FAQ question for more information on exact devices.

What devices are good for running LIQA in real-time?

We tested and optimized LIQA for the vast majority of modern mobile devices (a production year of 2018 or later). However, to achieve real-time performance (30+ FPS), we recommend using at least these hardware combinations:

Device hardware property

Minimal recommended setup

Mobile device RAM amount

3Gb

Mobile device GPU/CPU-paired

processor performance rate

Centurion Mark* > 100

What is a GPU Centurion Mark*?

GPU stands for graphics processing unit, and it is a specialized (usually additional) processor in modern digital devices for fast memory manipulation and parallel calculations. GPUs are widely utilized to accelerate image processing and neural network inference calculations.

Centurion Mark is a benchmark value calculated by the Tech Centurion company for comparison of mobile graphics processing units (GPUs). The benchmark involves both the automated computation of raw benchmarks and the manual testing of performance in games. As a result, it provides quite a good representation of mobile devices' real performance under high load during real-time image processing.

The latest Centurion Mark values for exact mobile GPUs are available on the Tech Centurion site: https://www.techcenturion.com/mobile-gpu-rankings#nbspnbspnbspnbspCenturion_Mark_for_Smartphone_GPUs

Some device examples:

Mobile phone

Production year

GPU/CPU-paired processor

Centurion Mark

LIQA can run with 30+ FPS?

Apple iPhone XR

2018

Apple A12 Bionic

119.3

Yes

Apple iPhone 12

2020

Apple A14 Bionic

134.8

Yes

Samsung Galaxy S20

2020

ARM Mali-G77 MP11 (Exynos 990)

114.7

Yes

Samsung Galaxy A51

2020

ARM Mali-G72 MP3 (Exynos 9610)

77

No

What technologies are required for LIQA work?

LIQA is a web-based tool that provides a better user experience across multiple platforms. LIQA uses JavaScript for all data and DOM manipulations. We also use common technologies for camera manipulation and algorithm optimization:

  • WebRTC for camera access, video and image capture;

  • WebGL for neural network optimization;

  • WebAssembly for algorithm optimization.

Is LIQA available as a React Native library?

Currently, this is not available, but soon LIQA Light React Native version will be available. Please, refer to LIQA Light documentation to get the latest updates

Is LIQA available as a library for native apps (e.g., Swift, Java, etc.)?

Currently, this is not available as a native library, but you can use LIQA Light through web-view technology. Please, refer to LIQA Light documentation.

How can I communicate if I have any issues with LIQA?

If you have not been previously contacted by a Haut.AI manager, please, let us know via the contact form on the haut.ai website. Describe the issue and leave an e-mail so we could reach you as soon as possible to help fix the issues.

Last updated