0
$\begingroup$

In the Tensorflow 2 Object Detection API model ZOO, there is 7 versions of the EfficientDet:

versions

Being new to deep learning, I don't understand very well what is the difference and how to choose the best one for my application (detecting buildings from satellite imagery). I haven't been able to find much info on this online.

Question: How do these differ and how does one decide which one to choose?

$\endgroup$

1 Answer 1

1
$\begingroup$

Each of the networks have different architectures. You can see the architectures of each on page 6 of the original paper.

Some rough generalizations are:

  • The lower numbered versions works best with low resolution images and the higher ones.
  • The lower ones are faster, but less accurate.
  • The lower ones requires less compute power than the higher ones.

So it's a trade-off that one will have to make depending on the use-case.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.