Mantis learns to classify 2D images for defect detection. Eventually, it should be able to handle other kinds of data and be easily configurable with a simple schema definition in a configuration file.
It is currently an early work in progress and is being tested on a bencmark dataset. It will have a GUI, TUI and CLI, with only the TUI using Textual being partially implemented currently.
Currently no Pypy package is provided and cloning the repo is required.
Installing with pip on *NIX OS:
cd /desired/install/directory
git clone http://github.com/jbergen/mantis
# Optionally activate a virtual environment here
pip install ./mantis
!TODO expand stub.
!TODO
!TODO
Main steps involved in the development of a Machine Learning Algorithm are:
Import dataset (Input and Output) | / Pre-processing (Filtering, upsizing/downsizing) | / Build model or fit regressions to data | / Vizualize Results Graphically and compare with validation
Here we see a linear regression Machine Learning Algorithm applied to a generalized data for determining quality of wine based on different parameters such as alcohol content, density, pH, etc.
. . .
Here, a machine learning algorithm which utilizes a Convolutional Neural Network is used to detect and classify defects on solar panels based on surface scratches on them.
. . .