Custom visualization of raw data

Help
#1

Hi,

I am trying to use Edge impulse UI for data different to audio or accellerometer. In my case from the raw data plot I cannot tell where I have actual signal and where I have just background. So for spitting the data I would require different type of visualization, or I would need to split the samples after custom preprocessing.

Can that be done in some way ?

Best Regards
step

#2

Hi @step22,

Please note that the visualizations in the UI are mainly set up for accelerometer, audio, and image data. The accelerometer plots can be used for nearly any time-series data, but they were originally configured for accelerometer data.

If the Edge Impulse UI plots do not help with visualizing your data, I might recommend finding another tool, such as Excel or Python with matplotlib (https://matplotlib.org/). Personally, I am a big fan of using Google Colab (https://research.google.com/colaboratory/) to curate, analyze, and plot my raw datasets before bringing them into Edge Impulse.

#3

Thanks for your answer.

One thing I particularly like about edge impulse is the ability to record and partition the dtaa through the webinterface … sure one can do that offline …
One could add custom visualization in the same way as custom procesing blocks by hooking in some http callbacks.