Awesome

nnio.l

A python package that you can use to create neural networks with one line of code.

Requirements:

Tensorflow==2.4.0
scikit-learn==0.24.0
opencv-python

Supported architectures:

1. Multilayer Perceptron [Image classification]
2. Convolutional Nerual Network [Image classification]

Dataset format:

Dataset
  |__LABEL 1
     |__IMG 1
     |__IMG 2
     .
     .
     |__IMG n
  |__LABEL 2
     |__IMG 1
     |__IMG 2
     .
     .
     |__IMG n
  .
  .
  .
  |__LABEL n
     |__IMG 1
     |__IMG 2
     .
     .
     |__IMG n

Example [Creating and training a new MLP]:

import nniol

nn = DenseNet(use_pretrained_model=False, path_of_dataset='<PATH OF DATASET HERE>', neurons_per_layer=[<LIST OF INTEGERS SPECIFYING THE NUMBER OF NEURONS IN EACH LAYER>], activations=[<LIST OF STRINGS SPECIFYING ACTIVATION FUNCTIONS FOR EACH LAYER>], model_path='<PATH TO SAVE MODEL>', epochs=<NUMBER OF EPOCHS TO TRAIN>)

nn.predict('<PATH OF DATA TO PASS FOR INFERENCE>')

Example [USING A PRETRAINED MLP]:

import nniol

nn = DenseNet(use_pretrained_model=True, model_path='<PATH OF SAVED MODEL>')
nn.predict('<PATH OF DATA TO PASS FOR INFERENCE>')

Example [Creating and training a new CNN]:

import nniol

nn = ConvNet(use_pretrained_model=False, path_of_dataset='<PATH OF DATASET HERE>', filters_per_layer=[<LIST OF INTEGERS SPECIFYING THE NUMBER OF FILTERS IN EACH LAYER>], activations=[<LIST OF STRINGS SPECIFYING ACTIVATION FUNCTIONS FOR EACH LAYER>], model_path='<PATH TO SAVE MODEL>', epochs=<NUMBER OF EPOCHS TO TRAIN>)

nn.predict('<PATH OF DATA TO PASS FOR INFERENCE>')

Example [USING A PRETRAINED CNN]:

import nniol

nn = ConvNet(use_pretrained_model=True, model_path='<PATH OF SAVED MODEL>')
nn.predict('<PATH OF DATA TO PASS FOR INFERENCE>')