ModiFace at CVPR 2024

June 21, 2024

We were thrilled to present two exciting papers at this year’s Conference on Computer Vision and Pattern Recognition #CVPR2024!

For our first paper, we explored dynamically predicting lighting maps of a scene from a single selfie image. We used synthetic data to train an AI model, and then further tested this on real world scenarios. Check out our paper “Inclusive Portrait Lighting Estimation Model Leveraging Graphic-Based Synthetic Data”. Authors: Kin Ching Lydia Chau, Tao Li, Ruowei Jiang, Zhi Yu, Panagiotis-Alexandros Bokaris, ModiFace Inc., L’Oréal Research & Innovation.

For our second paper on SCE-MAE: Selective Correspondence Enhancement with Masked Autoencoder for Self-Supervised Landmark Estimation we developed a groundbreaking framework for self-supervised landmark estimation that's blowing other techniques out of the water! Our novel framework outperforms existing state of the art methods by up to 1.4x accuracy on landmark matching. Authors: Kejia Yin, Varshanth Rao, Ruowei Jiang, Xudong Liu, Parham Aarabi, David B. Lindell, University of Toronto, ModiFace Inc.

We’re also super proud of our AI Lead, Ruowei Jiang, who was a speaker in a workshop presenting how we’ve achieved a lightweight realistic virtual try-on that runs in real time in a web browser.

Thank you to our presenters and authors for the continued innovation at ModiFace!