Real Time Image Saliency for Black Box Classifiers
This provides a fast, interpretable saliency method for real-time systems, but it is incremental as it builds on existing weakly supervised approaches.
The paper tackles the problem of real-time saliency detection for black-box image classifiers by training a masking model to manipulate classifier scores, achieving interpretable and artifact-free saliency maps that outperform other weakly supervised methods on ImageNet localization.
In this work we develop a fast saliency detection method that can be applied to any differentiable image classifier. We train a masking model to manipulate the scores of the classifier by masking salient parts of the input image. Our model generalises well to unseen images and requires a single forward pass to perform saliency detection, therefore suitable for use in real-time systems. We test our approach on CIFAR-10 and ImageNet datasets and show that the produced saliency maps are easily interpretable, sharp, and free of artifacts. We suggest a new metric for saliency and test our method on the ImageNet object localisation task. We achieve results outperforming other weakly supervised methods.