Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

FaceEraser: Removing Facial Parts for Augmented Reality

Sep 22, 2021
Miao Hua, Lijie Liu, Ziyang Cheng, Qian He, Bingchuan Li, Zili Yi

Share this with someone who'll enjoy it:

Our task is to remove all facial parts (e.g., eyebrows, eyes, mouth and nose), and then impose visual elements onto the ``blank'' face for augmented reality. Conventional object removal methods rely on image inpainting techniques (e.g., EdgeConnect, HiFill) that are trained in a self-supervised manner with randomly manipulated image pairs. Specifically, given a set of natural images, randomly masked images are used as inputs and the raw images are treated as ground truths. Whereas, this technique does not satisfy the requirements of facial parts removal, as it is hard to obtain ``ground-truth'' images with real ``blank'' faces. To address this issue, we propose a novel data generation technique to produce paired training data that well mimic the ``blank'' faces. In the mean time, we propose a novel network architecture for improved inpainting quality for our task. Finally, we demonstrate various face-oriented augmented reality applications on top of our facial parts removal model. Our method has been integrated into commercial products and its effectiveness has been verified with unconstrained user inputs. The source codes, pre-trained models and training data will be released for research purposes.

* 18 pages, 15 figures. ICCV 2022, Fifth Workshop on Computer Vision for AR/VR 

   Access Paper Source

Share this with someone who'll enjoy it: