I’m building a wearable band using the XIAO nRF52840 Sense.
The main goal is to implement tiny machine learning to identify the cause of forward head posture.
The classes will be divided into three categories: study (writing), smartphone use, and monitor viewing.
I’d like to utilize IMU sensor values for each activity.
Is it possible to use both frequency features and time-series data as training data in the free version?
You can combine multiple processing blocks as input into a single learning block in Edge Impulse. Just add multiple Processing Blocks to your Impulse in Impulse Design, then make sure you’re ticking both blocks as input into your Learning Block.
1 Like
Thank you for rply.
But in my project, each class mainly contains static values, so frequency analysis wouldn’t be meaningful, right? For example, when using smartphone, just keep head down.
potentially, though it’s worth considering even if the classes you’re trying to detect are static, your “all other movement” or “unknown” class may still have movement
Thank you for the quick reply 