These tools and metrics are designed to help AI actors develop and use trustworthy AI systems and applications that respect human rights and are fair, transparent, explainable, robust, secure and safe.
The learned perceptual image patch similarity (LPIPS) is used to judge the perceptual similarity between two images. LPIPS is computed with a model that is trained on a labeled dataset of human-judged perceptual similarity. The perception-measuring model computes similarity for arbitrary inputs by comparing activations of the model between two images of interest.
LPIPS can support the Robustness objective by providing a quantitative measure of how similar outputs remain under input perturbations or adversarial attacks. If an AI system's outputs are perceptually consistent (as measured by LPIPS) despite noise or minor changes, this indicates a degree of robustness. However, this connection is indirect and context-dependent, as LPIPS alone does not guarantee overall system robustness.
About the metric
You can click on the links to see the associated metrics
Objective(s):
Purpose(s):
Lifecycle stage(s):
Target users:
Risk management stage(s):
