Saliency-Benchmark

State-of-the-art Spectral Saliency Detection Models on state-of-the-art Datasets.

Installation

1. In Datasets/GroundTruth/ and Datasets/Images/, I only put two example images in each dataset folder. If you want to test on the real datasets, you need to replace them.
2. Run ./SaliencyBenchmark.m to generate saliency maps for all the models on all the datasets, the results are saved in ./SaliencyMaps/.
3. Run ./benchPR.m to save Precision and Recall results in ./Results/pr/.
4. Run ./benchPRF.m to save Precision, Recall and F-measure results in ./Results/prf/.
5. Run ./drawPR.m to show pr curves on all the datasets.
6. Run ./drawPRF.m to show histogram of Precision, Recall and F-measure results on all the datasets.