-1
\$\begingroup\$

I'm working on a project where I need to detect which finger touched the mobile screen i.e capacitive screen.

I'm not talking about double finger touch or single or drag and move etc. I'm asking which finger i.e middle finger, thumb, little finger, ring finger like this.

This is very helpful in drawing apps like I can draw something from my ring finger and erase from thumb. And I believe some other useful applications were there.

I have no idea on where to start. If anyone can help with any Idea I'm very glad.

\$\endgroup\$
23
  • 3
    \$\begingroup\$ I don't think this is possible at all but filming the hand & analyzing video un direct, but then never a mobile will have enough calculations/CPU performances to do that, I really think it is not possible that way. Second note : the hardware position of camera (not oriantable) won't permit a good capture do already two very major problems in the concept you're strting to think about. \$\endgroup\$
    – user201301
    Commented Nov 18, 2021 at 15:46
  • 1
    \$\begingroup\$ if I had to do that myself, I would try to use the camera never a sensor or pression differences or temperature differences & so on will be able to do a such specific identification or errors/fauls-positive on identification will be so high the application itself will be no-sense. Imagine an error level about more than 40% of cases.... \$\endgroup\$
    – user201301
    Commented Nov 18, 2021 at 15:51
  • 5
    \$\begingroup\$ I'm looking some kind of sensor which detects the fingers and identify which finger was that Then think about if such a sensor existed, how would it be able to identify the fingers? What physical principles would be used? It is easy to ask for a random sensor but that doesn't mean such a sensor is physically possible. Maybe you will need to sense something else and identify the finger from the data that is collected that way. \$\endgroup\$
    – Bimpelrekkie
    Commented Nov 18, 2021 at 15:59
  • 2
    \$\begingroup\$ It is interesting, but going to be a collection of opinions, I guess. IMO: Register every finger, fingerprint image recognition touch sensor needed. For in case, toe, the tip of the nose, and the tongue. \$\endgroup\$
    – jay
    Commented Nov 18, 2021 at 15:59
  • 2
    \$\begingroup\$ Different people have different shapes and different pressures applied on different fingers. Good luck researching on that with AI or ML. And whether all this research is worth its application is another question. \$\endgroup\$
    – Mitu Raj
    Commented Nov 18, 2021 at 17:34

1 Answer 1

Reset to default
3
\$\begingroup\$

If you want to do this with off-the-shelf hardware this will be difficult. Standard cell phones are not designed for this so you cannot use any of the existing sensors directly.

You may be able to leverage second order effects. You can collect large data bases of touch events with different fingers/people/devices etc. and then record duration, location, vibration, orientation, touch area over time, etc. Then throw a neutral network or suitable machine learning algorithm at it and see if can "learn" to distinguish the different fingers.

This is based on the hypothesis that different fingers generate different "fingerprints" (pun intended!) in the senor patterns and that a machine learning algorithm can actually correlate and latch on to. Example: you will have different wrist positions if you use your ring finger and maybe that's detectable in the gyro or accelerometer data.

IMO that's a long shot and will require significant amount of work, but I don't see any better alternative at the moment.

\$\endgroup\$
1
  • \$\begingroup\$ I believe the important point here is “and see if it can…”. The thumb can probably be differentiated in most cases. For the others… \$\endgroup\$
    – jcaron
    Commented Nov 18, 2021 at 22:55

Not the answer you're looking for? Browse other questions tagged or ask your own question.