Hi everyone,
What type of sensors are recorded from smartphones when collecting data directly from them? What 33 features are shown when I want to input these features to the neural network when collecting from motion sensors(smartphone)?
Hello @Bayan_khalid,
What kind of project are you using? considering the amount of features here I guess a movement recognition project correct? In that case we use the accelerometer from your phone and we collect the axis (accX, accY, accZ) for X seconds, depending on the frequency (probably 62.5 if you’re using your phone), you’ll get, let’s say for one second (1000ms), 16 values per axis (1000/62.5). So in total 48 values.
Then those “raw” values are passed in the pre-processing block to extract meaningful features. When you use the Spectral Analysis block, we use Fast Fourier Transforms to extract the features.
You can have more details here: Spectral features - Edge Impulse Documentation
If you’re not using an accelerometer project, feel free to navigate in the documentation to understand the pre-processing technique.
Best,
Louis
Louis forgot the link 
(@Louis Do you see “Sorry, we were unable to generate a preview for this web page, because the following oEmbed / OpenGraph tag could not be found: description”? Can the Team fix that so when we post links it does not appear?)
The project we are working on is related to car driving, we need a gyro sensor,
is there a way to enter it?