Urban Scene Understanding Example

CN24 ships with a very simple pre-trained model for urban scene understanding. For training, we used the LabelMeFacade dataset. You can use this model for fine-tuning as well, but in this short demonstration we will use it to classify images!

First, clone CN24 into a new directory. Make sure that libjpeg is installed on your machine and run the demo script from the source folder:

$ ./compile-and-test.sh

This script will compile CN24 and download a sample image from the eTRIMS dataset. If everything worked correctly, your should have a file output3.jpg in the build directory. Its contents should look like this: