but there are a lot of weird things going on...Like the ICA doesn't have many peaks, but when i pass just the green signal into the FFT it still outputs a similar spectrum and again about 50 bpm...
Yep, your output does not respond to your input. What you see is artifacts. I am about to head out for my trip, will not be able to respond (here) very quickly until tomorrow probably. Draw a sketch of your signal pipeline and follow it block for block. Input, output. What goes in, what comes out, what would you expect to come out. We'll be in touch
@TanMath quick note with some impact here. Keeping track of the frame timestamp is not about skipping frames. If you don't you will be designing your filters for specific frequencies but your signal will be in others because of the phase changes. Therefore, it might be good to check what goes in and out of the filter.
@A_A So we are not sure where the artifact is coming from? I will send you a block flowchart as well and test the signal processing pipeline... However, I am not sure how to test the computer vision part of the algorithm...
I guess there is not much to test there anyway because it is just measuring the RGB pixel values and averaging...
@TanMath First message timestamp is 05:49 UK time. I am in the UK. For artifact, I suspect, this is why I have been asking you questions rather than providing a definite answer. Not the computer vision. From acquiring average RGB per frame to estimating BPM. That part is what I am referring to. Flowchart would help
@A_A this reminds me, I saw an oscillatory signal in the background run that I did again...
lemme show you the plot...
@A_A hmmm problem is I can't seem to find a way to turn that off in opencv... There is something about setting the exposure, but I don't know what it would be set to... As you can see, I am a beginner with digital cameras