@viibeeng It’s pretty easy and maybe like $10. Use a guide like this and Dreambooth. Takes maybe 2 hours to collect your own images and then train.
https://t.co/mcmeostnY1
@aidangomezzz It's very exciting that you're equipped and willing to continue democratizing access to this tech for developers and organizations large and small!
@TGoesh@ReinH And this is what the code would look like:
if sigmoid(0.3*0.4, 0.22*0.6, …. <thousands, millions, or billions more operations>) > confidenceThreshold:
isFace = True
@TGoesh@ReinH Neural nets are already explainable in most cases with visualization and the weights would mean nothing to us depicted as code. Here’s a visualization of what triggers the “if statements” in the different layers of a deep neural net that is trained to detect faces in an image.