public EfficientNet — struct

EfficientNet(scalings, block_config;
             inchannels = 3, nclasses = 1000, max_width = 1280)

Create an EfficientNet model (reference). See also efficientnet.

Arguments

scalings: global width and depth scaling (given as a tuple)
block_config: configuration for each inverted residual block, given as a vector of tuples with elements:
- n: number of block repetitions (will be scaled by global depth scaling)
- k: kernel size
- s: kernel stride
- e: expansion ratio
- i: block input channels (will be scaled by global width scaling)
- o: block output channels (will be scaled by global width scaling)
inchannels: number of input channels
nclasses: number of output classes
max_width: maximum number of output channels before the fully connected classification blocks

EfficientNet(name::Symbol; pretrain = false)

Create an EfficientNet model (reference). See also efficientnet.

name: name of default configuration (can be :b0, :b1, :b2, :b3, :b4, :b5, :b6, :b7, :b8)
pretrain: set to true to load the pre-trained weights for ImageNet

Tutorials