Vislab

Vislab is a set of Python modules for visual recognition:

• Dataset interfaces (ImageNet, PASCAL VOC, AVA, our Flickr style, Pinterest style, Wikipaintings datasets).
• Distributed computation of image features.
• Training and testing predictors for classification and regression tasks.
• Evaluation of prediction results, with nice figures, confusion matrices, tables.
• Web UI for easy exploration of data and prediction results.

It is released under the BSD 2-clause license.

The github repository contains all source code and issue tracker.

Check out an image style classification demo!

Documentation

• Get started with the tutorial.
• Read about the provided datasets and how to add your own.

Acknowledgements

This code was written by Sergey Karayev while he was an intern at Adobe Creative Technologies Lab, a Graduate Student Researcher at UC Berkeley, and on his own time, and features contributions from others.