We use cookies to ensure that we give you the best experience on our website. If you continue to browse without changing your settings, we’ll assume that you’re happy to receive all the cookies on this Toshiba website. You can also click the “Continue” button to accept our policy in your browser so this message does not appear in the future. You can, of course, change your cookie preferences at any time via your browser settings. Details of our cookies and cookie policy are accessible under the link “Find out more”.

Continue Find out more

The Dataset

If you compare or use our network in your work, please cite the following paper:

ContextNet: Exploring Context and Detail for Semantic Segmentation in Real-time R. P. K. Poudel, U. Bonde, S. Liwicki and C. Zach
BMVC 2018 / arXiv

The network design and its training process is provided in the source files, and is to be used for research only. You will require Python, Tensorflow with GPU support and OpenCV2 to run the scripts. We use the Cityscapes dataset for training.

Download Tensorflow Model of ContextNet