transcendental 0.5.2

Minimal IA for image recognition

Transcendental (π)

Artificial Intelligence image recognition.

Learn from input data and generate model, predict from previous generated model

Disclaimer

This is testing project purpose so it's currently in development and features are minimal don't juge me :)

Features

• Recognize cats
• Recognize dogs

Install

$ pip install transcendental

Usage

trans-predict

This command allow you to recognize an image passed in argument.

Predict take exactly one argument who is the image to predict

$ trans-predict --help 
usage: trans-predict [-h] image

Recognize cat and dogs

positional arguments:
  image       an image to analyze

  optional arguments:
    -h, --help  show this help message and exit

$ trans-predict ./image.png

trans-train

This command allow you to train your IA.

This command generate models automaticaly used by predict at launch

Currently in development!

Requires

• python 2.7+
• keras
• tensorflow
• Pillow
• h5py

Resources

• https://becominghuman.ai/building-an-image-classifier-using-deep-learning-in-python-totally-from-a-beginners-perspective-be8dbaf22dd8
• https://drive.google.com/drive/folders/1XaFM8BJFligrqeQdE-_5Id0V_SubJAZe
• https://machinelearningmastery.com/save-load-keras-deep-learning-models/
• https://www.quora.com/How-do-artificial-neural-networks-work
• http://torch.ch/blog/2015/11/13/gan.html