Need Some Advice

[sayujyasanadhya0]

Hi!

First off, thank you for the post, I found it incredibly useful for my project.

I’m aiming to develop an accurate and low latency tracking solution for VR controllers which use surface EMG signals to detect hand position. While exploring techniques to turn the noisy EMG signals into accurate and readable data, I stumbled across your project on GitHub.

I was really excited by your approach to EMG processing in particular, especially the fact that it uses little computational resources and have a few questions.

If you had just a few minutes to offer me any brief advice on electrode placement, signal processing strategies, or common pitfalls in EMG-based hand tracking, I would greatly appreciate it. Even a short response would mean a lot to me.

Thank you so much for your time, as I'm sure you're quite busy. I truly admire the work you have done in this area.

[shayanalibhatti]

Hi, Thanks for finding this helpful. Since its been more than 5 years, i suggest not to use simple neural network approach. Instead, use LSTM or better approaches for robust and better classification. I am not sure if Myo is still in business or not, so look for any advancements in this sensor too. I remember the Myo service getting shut down some times, but i think code takes care of initiating it if its not working, as long as its installed. The armband has to be placed consistently at similar position, which might be hard to remember, as the position of armband will affect the data collection as pressure might change. So either train model by putting armband at different positions (which can be a hassle for customer) OR try to remember position. Also, you will have to experiment if a single model will work on different people e.g. thick, thin, different blood pressure or not. Or else, the model will work specific to each person only. So you will need to think about this. Also, why are you using this EMG armband if just hand position is to be tracked ? You dont need EMG, you just need accelerometer, gyroscope based sensor attached to hand. EMGs are not needed. So maybe cheaper options might be available in the market. Regards, Shayan

…

On Wed, Feb 25, 2026 at 12:51 PM sayujyasanadhya0 ***@***.***> wrote: *sayujyasanadhya0* created an issue (shayanalibhatti/Finger-Movement-Classification-via-Machine-Learning-using-EMG-Armband-for-3D-Printed-Robotic-Hand#4) <#4> Hi! First off, thank you for the post, I found it incredibly useful for my project. I’m aiming to develop an accurate and low latency tracking solution for VR controllers which use surface EMG signals to detect hand position. While exploring techniques to turn the noisy EMG signals into accurate and readable data, I stumbled across your project on GitHub. I was really excited by your approach to EMG processing in particular, especially the fact that it uses little computational resources and have a few questions. If you had just a few minutes to offer me any brief advice on electrode placement, signal processing strategies, or common pitfalls in EMG-based hand tracking, I would greatly appreciate it. Even a short response would mean a lot to me. Thank you so much for your time, as I'm sure you're quite busy. I truly admire the work you have done in this area. — Reply to this email directly, view it on GitHub <#4>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJY5TRM3DW72KJQW4ZK6KJL4NXVM5AVCNFSM6AAAAACV7UDCWOVHI2DSMVQWIX3LMV43ASLTON2WKOZTHE4TCMJVGE2DOMI> . You are receiving this because you are subscribed to this thread.Message ID: <shayanalibhatti/Finger-Movement-Classification-via-Machine-Learning-using-EMG-Armband-for-3D-Printed-Robotic-Hand/issues/4 @github.com>

[sayujyasanadhya0]

Hey!

Sorry, I meant finger position and not hand position, like gestures, extension etc

Thanks a lot for the advice! I will look into LSTM and try and keep the sensors in the same position. I guess I will also need to program a sort of calibration mode for each individual user too. I still need to research what sEMGs to buy because I don't have a lot of budget for this.

Again, thanks a lot for your time, this has been incredibly valuable for me.