nerajbobra / embedded_pedometer Goto Github PK

Hi, this algorithm looks like great, so I want to use it on our wearable watch product. Now I should run the algorithm by Python to show visual chart, but I am not sure the Python code I write is correctly, so can you attach the visual chart for the data you provide in the csv file?

Hello Team,

We have started implementing the step counting algorithm by using your algorithm.

From our IMU sensor we got the raw data and converted that into float data by following some sensitivity calculation.

Then, that decimal data has to be converted to int8 integers based on the comments. You have used matlab and calculated absolute and max then dividing with 127. As a result you got the scaling factor 55.3293. This value is then multiplying with decimal data.

If we use the scaling factor 55.3293 in our application then we are not meeting accuracy.

Can you please suggest how to calculate the scaling factor for our IMU sensor data, and mention the steps to do.

This information would be really helpful for our implementation.

Thanks in advance.

The SDE is Dev C++ on windows 7...

(1) While looking at the sample data, are the values normalized? Shouldn't one of the axis values be close to gravitation acceleration?
(2) What is the coordinate axis of the accelerometer with which the sample data was collected?
(3) How many steps were actually walked vs. detected using the algorithm?

Thanks!