/pixy

pixy: lightweight image training set creation for down-stream machine learning applications

Primary LanguageJupyter NotebookMIT LicenseMIT

About

Pixy is a lightweight application for generating image datasets intended for downstream machine vision applications. With a few image labeling functions, pixy is able to achieve near human accuracy for image turking. For more information about pixy, read "Computer Vision Through Perception: Semantic Understanding of Novel Scenes through Data Programming" (Lau and Guo, 2016).

Requirements

Pixy depends on DeepBox, pycaffe, caffe, snorkel, and vlfeat. Instructions to install each of these libraries can be found online. It is highly recommended to install these libraries on Ubuntu.

TODO:

Automate installation requirements and make a user-abstracted interface

Getting Started

Insert instructions here.

Acknowledgements

Insert acknowledgments here.