Hi, check in OpenMV IDE → Tools → Machine Vision → CNN library. It should be in there. You have to copy it to disk on the camera.
That said, that example only kinda works. The face detector is great, but, the smile detection CNN isn’t the most robust. The network isn’t quite large enough to be that solid.
Um, I’m working right now on getting tensor flow lite ported to the camera. I’ve gotten all the code compiling and it seems to run. However, I need to rig up an example now showing folks how to make a network step by step. I was hoping to have an ETA for the new feature this weekend but I have to get Google to optimize their tensor flow for microcontroller code a bit more before it will be ready.