keras-squeeze-excite-network
keras-squeeze-excite-network copied to clipboard
Pre-trained model
Does anyone have pre-trained weights available?
@cipher009 how to pre-trained the model? I check the keras/example/application and it show me how to pre-trained model. i check the se_densenet.py.
model = sedensenet121 (include_top = false,weights = 'imagenet')
but it show me an error.
If you check the code it actually never tries to load imagenet weights for most of the densenets.
Does anyone know if you can train these networks directly on ~50k images or do i need to get the imagenet dataset and 'burn it in' with this? Also, does anyone know how long it would take on !one! gtx1080ti?
Does anyone find the pre-trained weights?
The Se_ResNet script never actually loads weights.
Found weights for ResNet. https://github.com/qubvel/classification_models/blob/master/classification_models/weights.py
You will have to downgrade Keras version to 2.1.2 to run it.