To submit your model to the MIT/Tuebingen Saliency Benchmark, follow these steps:
Download the 300 images of the MIT300 test set or the 2000 images of the CAT2000 test set, depending on which benchmark track you intend to submit to..
Run your model to compute predictions for each model
Save the predictions to .mat, .npy, .jpg or .png files of the same size and name as the original image. If your model is probabilistic, save the log density predictions and don't use image files since they can't store the actual log density values. Please save them as a zip or tar folder with the name of your model as the folder name. Alternatively you can use pysaliency's "export_to_hdf5" function. Please avoid using rar files if possible.
Submit your maps to email@example.com. Please mention whether your model is a probabilistic model predicting fixation densities, or a classic saliency map model.
We post your score and model details on this page
Let us know if you have a publication, website, or publicly available code for your model that we can link to your score in results table.