An implementation of "NASNet" models from the paper Learning Transferable Architectures for Scalable Image Recognitio in Keras 2.0+.
Based on the models described in the TFSlim implementation and some modules from the TensorNets implementation
Weights have been ported over from the official NASNet Tensorflow repository.
Since CIFAR weights are not provided, and I don't have the resources to train such large models on CIFAR, those weights will not be provided. External help is appreciated.
All types of NASNet models can be built. In addition, NASNet Large - NASNet (6 @ 4032)
and NASNet Mobile - NASNet (4 @ 1056)
are prebuilt and provided as NASNetLarge
and NASNetMobile
.
from nasnet import NASNet
# the parameters for NASNetLarge
model = NASNet(input_shape=(331, 331, 3),
penultimate_filters=4032,
nb_blocks=6,
stem_filters=96,
skip_reduction=True,
use_auxilary_branch=False,
filters_multiplier=2,
dropout=0.5,
classes=1000)
from nasnet import NASNetLarge, NASNetMobile
model = NASNetLarge(input_shape=(331, 331, 3), dropout=0.5)