Suggested configurations

Introduction

Starting from version 3.18 of dasFace, the RAM requirement can be reduced drastically by customizing the models and/or endpoints to be used at runtime. In the next section, it can be found some of the suggested configurations according to the most common use cases. For new customers, using only our most modern biometric model besides all the antispoofing endpoints will at least halve the RAM memory usage of the service when running and serving requests.

Recommended configurations

Although the config.yaml can be modified, the following config.yaml are suggested for the different use cases.

Verification models	Photo antispoofing	Video antispoofing	RAM Usage per worker	File config
All models	True	True	12.3Gb	config.yaml
20221017	True	True	5.0Gb	config.yaml
False	True	False	3.8Gb	config.yaml
All models	False	False	10.4Gb	config.yaml
20221017	False	False	3.8Gb	config.yaml

Kindly take note that the RAM usage measurements provided have been conducted on a system equipped with an i7-9750H processor operating under Ubuntu 20.04. It is essential to notice that a variance of up to ±2GB may be encountered when utilizing dasFace under different operative systems and/or processors.

YAML schema

The config.yaml must follows the next schema:

    type: object
    additionalProperties: false
    properties:
        version: 1.0
        photo_antispoofing:
            type: boolean
        video_antispoofing:
            type: boolean
        verification:
            oneOf:
                - type: boolean
                - type: array
                    items:
                    pattern: ^[a-f0-9]{56,56}$
    required: [version, photo_antispoofing, video_antispoofing, verification]

Running in docker container

Here, there is an example of how to run a docker container using a configuration .yaml file.

docker run --name dasface \
--cap-add=CAP_SYS_CHROOT \
-e DASFACE_CONFIG_PATH=/path/to/config.yaml
-e WORKERS=2 \
-e LOG_LEVEL=WARNING \
--rm -t - i -p 8810:8810 \
-v /local/path/to/config.yaml:/path/to/config.yaml:ro \
registry.gitlab.com/dasgroup/veridas/face-team/products/dasface:3.18.4-onpremises-cpu