臉部匿名化簡單化

Face Anonymization Made Simple

November 1, 2024

作者: Han-Wei Kung, Tuomas Varanka, Sanjay Saha, Terence Sim, Nicu Sebe

cs.AI

摘要

目前的臉部匿名化技術常常依賴於臉部識別模型計算的身份損失，然而這可能不準確且不可靠。此外，許多方法需要額外的數據，如臉部標誌和遮罩來引導合成過程。相反地，我們的方法使用擴散模型僅具有重建損失，無需臉部標誌或遮罩，仍能生成具有精細細節的圖像。我們通過定量和定性評估在兩個公共基準上驗證了我們的結果。我們的模型在三個關鍵領域取得了最先進的性能：身份匿名化、臉部屬性保留和圖像質量。除了其主要的匿名化功能外，我們的模型還可以執行臉部交換任務，通過將額外的面部圖像作為輸入，展示了其多功能性和多樣應用的潛力。我們的程式碼和模型可在 https://github.com/hanweikung/face_anon_simple 下載。

English

Current face anonymization techniques often depend on identity loss calculated by face recognition models, which can be inaccurate and unreliable. Additionally, many methods require supplementary data such as facial landmarks and masks to guide the synthesis process. In contrast, our approach uses diffusion models with only a reconstruction loss, eliminating the need for facial landmarks or masks while still producing images with intricate, fine-grained details. We validated our results on two public benchmarks through both quantitative and qualitative evaluations. Our model achieves state-of-the-art performance in three key areas: identity anonymization, facial attribute preservation, and image quality. Beyond its primary function of anonymization, our model can also perform face swapping tasks by incorporating an additional facial image as input, demonstrating its versatility and potential for diverse applications. Our code and models are available at https://github.com/hanweikung/face_anon_simple .