Move a mouse with its head

[PiGraham]

Setting aside the camera idea for a bit, if a solid state gyroscope or accelerometer is sensitive enough, it would have the advantage of being easier to program. However I do not know how subtle or small movements those boards are designed for. For sure, I²C is easy to work with in Python when the device in question is supported. The data streams would be simpler than a camera feed. The accelerometer boards look small and if they are like the other boards of similar make they would fit nicely on any head band or hat, since a hat or headband seems to be part of the requirements. The disadvantage there would be that a slim cable would be needed to connect it to either a full Raspberry Pi or a Pico. Though no soldering would be required for the models using Qwiic connectors. Perhaps someone here more familiar with the Pico could say authoritatively but I would even speculate that a Pico could pass the accelerometer readings on to a main computer and give the main computer the illusion that a mouse is connected.

I think you are right.

There are media remote controllers known as 'air mouse' that seem to use IMUs to detect motion and control a pointer.
The remote for an LG OLED TV works this way. It is very effective.
Devices are available that are compatible with Mac so it would be easy to try it out.

The disadvantage could be that it is an active device on the head so it needs wires or batteries. Off the shelf devices might not be comfortable or discrete.

A custom design could be much smaller. even invisible at the back of the head. The size of the remote control type is chosen to fit the hand.

[ame]

Setting aside the camera idea for a bit, if a solid state gyroscope or accelerometer is sensitive enough, it would have the advantage of being easier to program. However I do not know how subtle or small movements those boards are designed for. For sure, I²C is easy to work with in Python when the device in question is supported. The data streams would be simpler than a camera feed. The accelerometer boards look small and if they are like the other boards of similar make they would fit nicely on any head band or hat, since a hat or headband seems to be part of the requirements. The disadvantage there would be that a slim cable would be needed to connect it to either a full Raspberry Pi or a Pico. Though no soldering would be required for the models using Qwiic connectors. Perhaps someone here more familiar with the Pico could say authoritatively but I would even speculate that a Pico could pass the accelerometer readings on to a main computer and give the main computer the illusion that a mouse is connected.

I think you are right.

There are media remote controllers known as 'air mouse' that seem to use IMUs to detect motion and control a pointer.
The remote for an LG OLED TV works this way. It is very effective.
Devices are available that are compatible with Mac so it would be easy to try it out.

The disadvantage could be that it is an active device on the head so it needs wires or batteries. Off the shelf devices might not be comfortable or discrete.

A custom design could be much smaller. even invisible at the back of the head. The size of the remote control type is chosen to fit the hand.

Could you post a link to such a project? Somebody must have made one.

[magurp244]

Hm, if you have a bit of cash lying around I might also suggest something like the OpenBCI + cEEGrids. Although they also have dry contact headsets available with varying channels/price ranges, can also get it a bit cheaper i you can 3D print your own frame.

[ame]

How about no tracking dot? Just use a webcam to detect the user's face and track movement. Here's an example which could be extrapolated to this application:

https://hackaday.com/2023/03/12/immersi ... le-webcam/

[PiGraham]

How about no tracking dot? Just use a webcam to detect the user's face and track movement. Here's an example which could be extrapolated to this application:

https://hackaday.com/2023/03/12/immersi ... le-webcam/

Apparently that wont do for foufouner :

OpenCV could be a good option

https://pyimagesearch.com/2019/04/01/pa ... nd-opencv/

I would prefer an object rather than the face. I'll explain why: when you're in front of the computer, it will always recognize the face. Whereas when we are in front of the computer, but without the object, it will not recognize us.

Suggestions that tacking of head, eyes, marker etc) could be enabled and disabled as needed were dismissed.
I'm not sure why.
The request seems to be to make a copy of Tracker Pro. and nothing else.

[foufouner]

The request seems to be to make a copy of Tracker Pro. and nothing else.

Exactly

[ame]

The request seems to be to make a copy of Tracker Pro. and nothing else.

Exactly

Ok. Well I posted a link to a list of such projects at GitHub. But I would speculate that Tracker Pro is probably quite simple, and its fundamental operations could be described precisely and re-implemented easily.

[foufouner]

Here are the characteristics of the dots.
Which fire camera to cap