The ObjectProposal class in .layers.objection_detection is the one causing this error.
This seems to be because of the upgrade of Keras from 2.1.6 to 2.2.
A lot of the online chatter says to just change all instances of topology to saving, that doesn't seem to work in your code.
Is there a solution that doesn't involve downgrading Keras?
Temporary Fix:
Replace keras.engine.topology.Layer with keras.engine.base_layer.Layer, I don't know if this is how they're going to want it from now on, but for now it works.
The
ObjectProposal
class in.layers.objection_detection
is the one causing this error. This seems to be because of the upgrade of Keras from 2.1.6 to 2.2.A lot of the online chatter says to just change all instances of
topology
tosaving
, that doesn't seem to work in your code.Is there a solution that doesn't involve downgrading Keras?
Temporary Fix: Replace
keras.engine.topology.Layer
withkeras.engine.base_layer.Layer
, I don't know if this is how they're going to want it from now on, but for now it works.