API Reference: Classification Models

TensorNets provides a rich collection of pre-trained models for image classification. All models share a consistent functional API.

Common Signature

Each classification model follows this general signature:

def ModelName(inputs, is_training=False, classes=1000, stem=False, scope=None, reuse=None):

inputs: A tf.Tensor representing the batch of input images.
is_training: A boolean Tensor or a Python boolean. Controls behavior for layers like Batch Normalization and Dropout. Defaults to False.
classes: The number of output classes for the final layer. Defaults to 1000 for ImageNet.
stem: If True, the model returns the output of the convolutional base (the feature extractor) before the final classification layer. This is useful for transfer learning. Defaults to False.
scope: A string for the tf.variable_scope.
reuse: A boolean for reusing variables within the scope.

Below is a list of the available image classification models, grouped by their architectural family.

Model Family	Model Name	Source File
CapsuleNet	`CapsuleNet`	`tensornets/capsulenets.py`
Darknet	`Darknet19`	`tensornets/darknets.py`
	`TinyDarknet19`	`tensornets/darknets.py`
DenseNet	`DenseNet121`	`tensornets/densenets.py`
	`DenseNet169`	`tensornets/densenets.py`
	`DenseNet201`	`tensornets/densenets.py`
EfficientNet	`EfficientNetB0` - `B7`	`tensornets/efficientnets.py`
Inception	`Inception1` (GoogLeNet)	`tensornets/inceptions.py`
	`Inception2`	`tensornets/inceptions.py`
	`Inception3`	`tensornets/inceptions.py`
	`Inception4`	`tensornets/inceptions.py`
	`InceptionResNet2`	`tensornets/inceptions.py`
MobileNet	`MobileNet25`, `50`, `75`, `100`	`tensornets/mobilenets.py`
	`MobileNet35v2` - `140v2`	`tensornets/mobilenets.py`
	`MobileNet75v3large`, `100v3large`	`tensornets/mobilenets.py`
	`MobileNet75v3small`, `100v3small`	`tensornets/mobilenets.py`
NASNet	`NASNetAlarge`	`tensornets/nasnets.py`
	`NASNetAmobile`	`tensornets/nasnets.py`
	`PNASNetlarge`	`tensornets/nasnets.py`
ResNet	`ResNet50`, `101`, `152`	`tensornets/resnets.py`
	`ResNet50v2`, `101v2`, `152v2`, `200v2`	`tensornets/resnets.py`
ResNeXt	`ResNeXt50`, `101`, `50c32`, `101c32`, `101c64`	`tensornets/resnets.py`
SqueezeNet	`SqueezeNet`	`tensornets/squeezenets.py`
VGG	`VGG16`, `VGG19`	`tensornets/vggs.py`
WaveNet	`WaveNet`	`tensornets/wavenets.py`
WideResNet	`WideResNet50`	`tensornets/resnets.py`
ZFNet	`ZF`	`tensornets/zf.py`