How can we upload non-timeseries data using ingestion service? The binary attachments can be used for jpeg image but for a simple float32 1D or 2D array what is the best way to upload?
We don’t support non-timeseries data in Edge Impulse at the moment (except for images). You can create artificial timeseries data if you have a 1D array for example, just set the interval to 10ms. or so and ignore it otherwise.
Thanks @janjongboom for the suggestion! How should the payload look like?
I tried with the below data acquisition format but does not work.
"payload": {
"interval_ms": 10,
"sensors": [
{
"name": "temperature",
"units": "Cel"
}
],
"values": [26.0,26.0,25.5,25.0,25.6,25.8,25.0,25.3]
}
Also, I tried with
“values”: [“26.0,26.0,25.5,25.0,25.6,25.8,25.0,25.3”]
“values”: “26.0,26.0,25.5,25.0,25.6,25.8,25.0,25.3”
But it gets failed to upload using the edge-impulse-uploader.
EDIT: Since it should be a time-series (fake) so 2D array worked!
“values”: [[26.0],[26.0],[25.5],[25.0],[25.6],[25.8],[25.0],[25.3]]
1 Like