Sorry if this is a noob question. I have found your tuto on Continuous motion recognition very interesting and I am wondering if you have any idea/figure on how much power is used to run the model on the device like say for 1 minute. The bottom line is I am wondering if this solution would be viable in my use case where I would have to run the model for hours. Any help/hint would be greatly appreciated.
Hi @binbag, it’s really dependent on much you do on the device. You can sample data very efficiently (running the device at a low clock speed), then switch to a higher speed to do the inferencing, and go back to sleep. I have some info on power management and measuring this on development boards here: power management on Mbed OS.