Explainable Face Verification

Why does a model verify a person?

This plattform provides model-agnostic explanations for face verification and visually illustrates models decisions on the LFW, CALFW, CPLFW, SLLFW, and XQLFW database.

Method

How do we generate model-agnostic Explanations?


Systematic image occlusion
Facial feature extraction
Patch-wise distance deviation
Mapping and Color Blending

Viewer

Analyze Image Pairs and see Explanations and Metadata

Usage

Please cite our work


Explainable Model-Agnostic Similarity and Confidence in Face Verification PDF

            @inproceedings{knoche2023explainable,
              title={Explainable Model-Agnostic Similarity and Confidence in Face Verification},
              author={Knoche, Martin and Teepe, Torben and H{\"o}rmann, Stefan and Rigoll, Gerhard},
              booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
              pages={711--718},
              year={2023}
              }