While its been proven a fair few times that with a well crafted photo you can enter most systems with facial recognition. For a personal PC there is something kinda of nice about looking at the webcam unlocking your login session, lock screen or providing sudo access.

Using howdy this is done using very little user tweaking of things like pam.

This will work on most modern linux systems and the guys github page points you in the right direction for most distros

boltgolt/howdy

🛡️ Windows Hello™ style facial authentication for Linux – boltgolt/howdy

GitHubboltgolt

For this guide however I will focus on Ubuntu 20.04.1 Desktop 64bit.

Some things to be aware of..

• All the data for this photos etc is stored locally
• Howdy is built using OpenCV and Python
• The installation is going to take about 600Mb of space on your disk
• Facial recognition isn’t 100% secure and should never be your primary method of access on a production system.

Installing Howdy

Howdy is installed via an Ubuntu PPA

sudo add-apt-repository ppa:boltgolt/howdy

sudo apt update

sudo apt install howdy

The installation isn’t your normal sit back and you’re done install, its interactive and may look like its erroring out some times.

There is a question about how you wish Howdy to perform with the answers being

• Fast – The facial matches are more “fuzzy” however is quicker
• Balanced – Still quite quick, however if you’re further away may not log you in
• Secure – As the name suggests it is secure however will be slower

I chose Balanced and have had good results so far

You may also get what look like errors as what I assume is a make file or a python installer working that look like this

This seems to be ok

The whole installation takes about 5 minutes on my Dell XPS 13 (2019)

There may be questions about your webcam, and I just said yes to the defaults as we can edit that once Howdy is installed.

Once Complete its time to setup that webcam

Setup your webcam

Howdy uses some IR detection and needs both Video Capture AND Metadata capture on the webcam

We can find out more about the webcam by installing v4l-utils

sudo apt install v4l-utils

Once installed you will want to list the video devices on the system

v4l2-ctl --list-devices

This should return something like this

Integrated_Webcam_HD: Integrate (usb-0000:00:14.0-5):    /dev/video0    /dev/video1

Which is interesting because I only have one webcam on the device

Remember we needed to find out if the webcam did both video and metadata caputure, we can  use the following command

v4l2-ctl --device=/dev/video0 --all | grep Capture

This will return something like this

Video CaptureMetadata CaptureVideo CaptureFormat Video Capture:Crop Capability Video Capture:Selection Video Capture: crop_default, Left 0, Top 0, Width 640, Height 480, Flags: Selection Video Capture: crop_bounds, Left 0, Top 0, Width 640, Height 480, Flags: Streaming Parameters Video Capture:

Whoch would indicate the camera supports BOTH capture modes, specifically the first two lines

Video CaptureMetadata CaptureVideo Capture

For reference running the  command without the grep returns a lot of data about your video camera

v4l2-ctl --device=/dev/video0 --all

On 99.9% of systems /dev/video0 should be the right webcam

Update the Howdy Config

Once you know the location of your video device, the Howdy config needs to be edited

There are two ways of doing this

sudo howdy config

or

sudo nano  /lib/security/howdy/config.ini

The former basically runs the latter, both will open the config file in nano

Run a search for device_path and edit it to look like this

# The path of the device to capture frames from# Should be set automatically by an installer if your distro has onedevice_path = /dev/video0

Save and exit the file

Other Config Options

Well I say save and exit, however there are other options in this config file some which may help debugging if you have issues later. Some examples include

# The certainty of the detected face belonging to the user of the account# On a scale from 1 to 10, values above 5 are not recommended# Lower is bettercertainty = 2.8

# Disable Howdy in remote shellsignore_ssh = true

# The number of seconds to search before timing outtimeout = 4

Its worth having a look at the other options.

Provide your Face Images

The software is installed, the webcam is setup next it is time to add your face

sudo howdy add

You’ll run the add command, look at the webcam, if the webcam light goes on for a few seconds then off, you have sucessfully added a new Face profile, you can add as many of these as you want.

NOTICE: Each additional model slows down the face recognition engine slightlyPress Ctrl+C to cancelAdding face model for the user davidEnter a label for this new model [Model #1] (max 24 characters)

Add a name, and pose looking at the camera

Please look straight into the cameraScan completeAdded a new model to david

Use the following command to list the face models you have

sudo howdy list

This should return someting like this

Known face models for david:    ID  Date                 Label    0   2020-09-01 17:24:05  face1    1   2020-09-01 17:24:22  face2    2   2020-09-01 17:24:34  face3    3   2020-09-01 17:26:12  face side    4   2020-09-01 19:51:50  slightly darker room

Should you want to remove one run the command

sudo howdy remove <ID>Example: sudo howdy remove 3

This would remove “face side”

To clear all the ID’s run

sudo howdy clear

Take 2 or 3 Face photos closer, further away, bright light, darker room etc

Smile

Its time to test this out. The first test is checking Sudo access works

Open a new Terminal/Terminal Tab

type

sudo howdy list

Watch the webcam light, it should go on, for about 4 seconds and then display

Identified face as davidKnown face models for david:    ID  Date                 Label    0   2020-09-01 17:24:05  face1    1   2020-09-01 17:24:22  face2    2   2020-09-01 17:24:34  face3    3   2020-09-01 17:26:12  face side    4   2020-09-01 19:51:50  slightly darker room

The next test is the Lock screen

Click on Lock to lock the screen

Then press space to unlock

Again the webcam should go on for 4 seconds and unlock the screen

Finally test the Login screen, reboot and try and login

After the webcam light goes on, 4 seconds and you should login

You’re done. Facial recognition is enabled on your Ubuntu install

What does the installer change?

The Pam Module /etc/pam.d/pam-auth has a line added

auth    [success=2 default=ignore]        pam_python.so /lib/security/howdy/pam.py

Which has the PAM module call the python script each time authentication needs to take place.

Other Howdy Commands

The installer adds a howdy command to manage face models for the current user. Use howdy --help or man howdy to list the available options.

Usage:

howdy [-U user] [-y] command [argument]

• add – Add a new face model for an user
• clear – Remove all face models for an user
• config – Open the config file in your default editor
• disable – Disable or enable howdy
• list – List all saved face models for an user
• remove – Remove a specific model for an user
• snapshot – Take a snapshot of your camera input
• test – Test the camera and recognition methods
• version – Print the current version number

Conclusion

While I would never recommend this on a production system, I do advocate the usage from a pure laziness factor. Because of this I’m now going to figure out how to setup proximity locking on Ubuntu.

On the XPS13 i7 the results are quick and so far in a home situation this has worked really well.

References

How to Set Up Facial Recognition to Sign into Ubuntu and Other Linux Distributions

Missing the Windows Hello kind of unlocking of your system on Linux? Learn how to set up face unlock on Ubuntu and other Linux distributions.

It’s FOSSAbhishek Prakash

boltgolt/howdy

🛡️ Windows Hello™ style facial authentication for Linux – boltgolt/howdy

GitHubboltgolt