Скачать или смотреть Quantitative Metrics for Evaluating Explanations of Video DeepFake Detectors (BMVC 2022)

Authors:
Federico Baldassarre, Quentin Debard, Gonzalo Fiz Pontiveros, Tri Kurniawan Wijaya

Abstract:
The proliferation of DeepFake technology is a rising challenge in today's society, owing to more powerful and accessible generation methods. To counter this, the research community has developed detectors of ever-increasing accuracy. However, the ability to explain the decisions of such models to users is lacking behind and is considered an accessory in large-scale benchmarks, despite being a crucial requirement for the correct deployment of automated tools for content moderation. We attribute the issue to the reliance on qualitative comparisons and the lack of established metrics. We describe a simple set of metrics to evaluate the visual quality and informativeness of explanations of video \df classifiers from a human-centric perspective. With these metrics, we compare common approaches to improve explanation quality and discuss their effect on both classification and explanation performance on the recent DFDC and DFD datasets.

Website: https://github.com/baldassarreFe/deep...

Paper: https://arxiv.org/abs/2210.03683

Presented at the British Machine Vision Conference (BMVC 2022)