×

4D Light Field Benchmark

Check out the new Google group to connect with other light field researchers, discuss ideas, and initiate collaborations. Some questions about the benchmark are answered over there as well! There is also a new collection of light field resources. Please feel encouraged to add further links.

Please visit the repositories on our project page on github to checkout the latest version of the various light field tools: https://github.com/lightfield-analysis.

Import/Export Scripts

To easily operate with our data, we provide Matlab and Python tools for file import/export. Please visit the respective repositories for further details.

We further recommend the light-weight cvkit developed by Heiko Hirschmüller. It is a collection of visualization and conversion tools designed to work with PFM files. The official versions for Linux and Windows can be downloaded from the Middlebury benchmark page: http://vision.middlebury.edu/stereo/code/

Evaluation Toolkit

We provide an evaluation package with detailed submission instructions and source code of the evaluation toolkit (Python 2.7). You may use the toolkit to compute metric scores, validate submissions, and create figures. We further provide algorithm disparity maps for the training and stratified scenes (benchmark snapshot on September 19th 2019).

Blender Addon

Feel free to download our blender addon to create your own light field data. We are looking forward to see/hear about the awesome datasets that you create :)

blender-addon on github