Question about upload datasets

Hello, and I would like to ask if I can do the training in the Adge Impulse based on a column (label) in the Daset that was uploaded.

Thank you

Hi @Bayan_khalid,

I’m afraid I am not following your question. You can upload data in a variety of ways (see this page for more information: Uploader - Edge Impulse Documentation).

If you are uploading data, it should not contain the ground truth label of your sample (i.e. in any columns). The label (ground truth) must be set either in Edge Impulse, as part of the title of the file (e.g. label.sample1026.csv), or as the x-label if using the ingestion API (Ingestion API - Edge Impulse API). Hope that helps!

My question is, in particular, if these are the contents of the data set that I uploaded, and the label represents 0 top 1 bottom 2 right. I want it to do the training based on distinguishing between them. Can I do that?

Can I do that or not?

Hi @Bayan_khalid,

Yes. You will likely need to perform data standardization or normalization before uploading to Edge Impulse and use a Raw DSP block with a neural network classifier.