How can we help you today?

Why can’t I select different ODRs for the Accelerometer and Gyroscope?

The accelerometer and gyroscope are on a single Inertial Measurement Unit (6-axis IMU), and currently share settings.

What’s the maximum data amount I could collect?

Currently, Bronze users can collect up to 12 hours of data per Collection.

What is the highlighted region on the visualized signal displayed during “Event” data collection?

Qeexo AutoML automatically highlights the most informative part of the signal where the event is localized. These localized signals are further processed through the ML model-building framework to achieve the best classification performance.

What is ODR and which ODR should I choose?

Please refer to our blog titled: ODR and FSR of Sensors.

What is Full Scale Range and which FSR should I choose?

Please refer to our blog titled: ODR and FSR of Sensors.

What causes data check warnings?

Sensors and sampling rate mismatches, missing/duplicate/invalid values, and the saturation of signals can all cause data check warning. For more information, please refer to the AutoML User Guide.

What causes data a check error?

There may be a connection issue to the server, causing data check to time out.

Should I collect “Continuous” or “Event” data?

That depends on you use case and the pattern of your data. Think of it as excitations on the sensor streams: if your event of interest happens once in a while and have a pattern of finite duration (signal appearing for a short time window and phase out), select “Event” type; if your event of interest is continuous (constant signal that may or may not fluctuate), select “Continuous” type. An example of an “Event” type data is knocking on a surface with a mounted accelerometer.

How can I know whether the data is good or not from data visualization?

If you are collecting “Event” type data, you should inspect the highlighted regions on the graph that shows up immediately after each recording, and see whether the Events are properly localized and distinguishable from the rest of the signals. If you are collecting “Continuous” type data, you want to visually check the labeled data stream is distinguishable from the Noise/Background data (which you should have also collected). You may want to check sensor specific signals on different axes of the multi-channel sensors and compare the against your domain knowledge.

Still need help?