Using Tensorflow object detection API vs Keras
-
09-12-2020 - |
Question
I am new to machine learning. I am curious to know what is the difference between using Keras instead of TensorFlow object detection API. We need to manually configure hidden layers and input layer in Keras so what is the advantage to use Keras and how to know how many layers should configure to achieve object detection using Keras.
Please check two different types of implementation 1) Using Keras 2) Using Tensorflow Object detection API without Keras
Thanks !!!
Solution
Keras provides you high level api or can say wrapper written on top of multiple backends. These back ends have the core implementation of DNN. List of Keras supported backends are:
- Tensorflow
- Theano
- CNTK
**Source: Keras documentation for supported backends
Keras hides a bit complexity of DNN implementation, but again restrict your freedom. In case if you write a code in Tensorflow, you have explicitly specifies and calculate optimizer, cost function and other things, but it provides you flexibility.
So for me writing in Keras just a convenience.
As far my knowledge is concern, so far we dont have any fix formula to identify number of layers sufficient for Object detection :).