cancel
Showing results for 
Search instead for 
Did you mean: 

Biometric Authentification of a Person using Behavioral data from LSM6DSOTR

MShaf.1
Associate

Hello,

Is it possible to uniquely identify a person using data generated by the wrist/arm gesture/movements from LSM6DSOTR installed as part of a wrist wearable?

Does STM provide any SDKs to identify a person based on this data?

Regards

Shafi

1 ACCEPTED SOLUTION

Accepted Solutions
Eleon BORLINI
ST Employee

Hi @MShaf.1​ ,

This is -I believe- an high-level application, that might involve medical or security themes that are outside ST purposes.

There are of course a lot of examples on pattern recognition for gesture and activity powered by ST sensors and with the LSM6DSO(X), for example you can refer to this Github repository, where examples of Activity recognition, Head gestures or Gym and Yoga poses are available, exploiting the MLC core of the LSM6DSOX.

Probably, combining the info coming from a single motion sensors, it is possible to draw a unique pattern for each person.

Of course, if you could use other sensors such as a microphone, it would be easier to construct a unique pattern for single persons.

Another approach to this topic is to use "low-level" existing solutions such as already available design kits (for example, the SensorTile.box equipped with the LSM6DSO(X)) and the related SDKs (the ST BLE Sensor, available on Android and iOS stores and also in source code on Github:( this will facilitate the raw data acquisition and elaboration, and for more advanced pattern recognition purposes you can use together the Algobuilder tool to make the algorithm building and the code porting on the tool easier.

It's not a trivial topic, but if my reply answered -at least partially- your question, please click on Select as Best at the bottom of this post. This will help other users with the same issue to find the answer faster. 

-Eleon

View solution in original post

2 REPLIES 2
Eleon BORLINI
ST Employee

Hi @MShaf.1​ ,

This is -I believe- an high-level application, that might involve medical or security themes that are outside ST purposes.

There are of course a lot of examples on pattern recognition for gesture and activity powered by ST sensors and with the LSM6DSO(X), for example you can refer to this Github repository, where examples of Activity recognition, Head gestures or Gym and Yoga poses are available, exploiting the MLC core of the LSM6DSOX.

Probably, combining the info coming from a single motion sensors, it is possible to draw a unique pattern for each person.

Of course, if you could use other sensors such as a microphone, it would be easier to construct a unique pattern for single persons.

Another approach to this topic is to use "low-level" existing solutions such as already available design kits (for example, the SensorTile.box equipped with the LSM6DSO(X)) and the related SDKs (the ST BLE Sensor, available on Android and iOS stores and also in source code on Github:( this will facilitate the raw data acquisition and elaboration, and for more advanced pattern recognition purposes you can use together the Algobuilder tool to make the algorithm building and the code porting on the tool easier.

It's not a trivial topic, but if my reply answered -at least partially- your question, please click on Select as Best at the bottom of this post. This will help other users with the same issue to find the answer faster. 

-Eleon

MShaf.1
Associate

Hi @Eleon BORLINI​ ,

Thanks for your detailed response and Github link. We have the following sensors on a healthcare wristband

a) accelerometer + gyroscope

b) ECG sensor / Heartrate sensor

c) body temperature sensure

Based on these sensor data, we will be creating a unique pattern for a single person.

Regards

Shafi