Well that's not really an Arduino question is it: that's a question to do more with how your model works. If you trained it from internet data from 2 sensors, sounds to me like you need to just pass the data from both your actual sensors through to the model rather than combine them into some kind of "hybrid". But if you do want to pass some combination to the model, surely the make up of that combination is something you need to know before coming here, perhaps to ask about how to code the Arduino to do that combination?
edit: Which is pretty much what @Idahowalker said a while back:
I guess my question is simple and clear i don't know why you confused . if you have a reading signals coming from two analog inputs form the arduino ,how to combine the 2 signals and make it as one input . should i use the avarage value of the 2 sensor or i use the defferential of the 2 signals or there is another way to do that
And I'm saying, that's not for us to say, it's for you to know as the user of some model. To do either of those things is trivial: add them up and divide by 2, or take the absolute value of one minus the other. Or a third way might be a geometric mean, the square root of the product.
How would an Arduino forum know (other than by sheer fluke of having an expert on your model here) what to do with your data to condition it for some model?
But answer this please: what makes you think that you need to combine the data? If your model was trained with data from 2 sensors, surely it needs both data to run?
How did you train the model to use both combined signals?
You'd want to combine the signals the same way you combined the signals to get the trained model. If your trained model does not use combined signals then combining the signals would NOT produce useable results.
I use KNN/CNN to detect animals. I use an Infared camera to take pictures which are then converted into an input for a trained model.
I took 100's of images of animals and used those images to create trained models. My experience shows that it is best to use a trained model that is based upon the sensors 'final' groomed output. Trying to groom an input to trained model's to fit the trained model will not result in good results.
Did you train the model to use a combination of 2 sensors?
And putting -9V into a Uno's analog pin will not work.
Dude the model is not the issue whatever model will result the same thing the concept of the model is the same .The thing is after you train the model and you want to use the model that the issue.Yor case of using the image is different since you are using one camera
Did you train the model with a combined sensor reading? Combine the signals the same way the signals were combined to train the model, right?
Actually, I am using 2 Infared cameras that combine the image into one image to send to the KNN engine.
Dude, each camera starts at pixel 0, I sum the pixel information from both cameras to get a single image that is then processed by KNN. I trained my KNN model using a summed image from 2 cameras.
You need a reference point, a sync point. The sync point I am using is pixel 0 from both cameras. A sync point is needed to combine the readings. What will be the sync point of the two EMG sensors?
Use the same setup that was used to create the trained model that uses the combined data from both sensors.
That is absolutely closed to my inquiry for your case you used two cameras and combined it to be one.The same thing come to mine but the only thing is that I'm using the analog input of the Arduino