About

Fatemeh works as Early Stage Researcher at Technische Universität Berlin (TUB), Germany, and her research topic is “Headset Removal for Gaze Contact in XR Applications“.

You can read Fatemeh’s introduction here.

Fatemeh created and recorded a Science Slam on her research topic, which you can watch here

HEADSET Release

HEADSET (HEADSET: Human Emotion Awareness under Partial Occlusions Multimodal DataSET) and its all associated raw data and license agreement are now publicly available for research purposes. Please check the following link to find all the details: https://webpages.tuni.fi/headset/ 

Research Focus    

Fatemeh has been concentrating on the head-mounted display (HMD) removal problem in extended-reality (XR) applications. As a potential outcome of her research project, virtual communication can be improved by recovering the missing image information caused by the occlusion of an HMD in a realistic manner. Fatemeh collected a multimodal dataset (including multiview RGB-D, textured mesh, point cloud, and light field) through user experiments during her secondment at Tampere University (CIVIT Lab) to understand the usage of high-dimensional and real-world data in solving her research problem. She is also working on video inpainting algorithms designed based on generative adversarial networks (GANs) and applied to the collected data in XR telecommunication scenarios. Furthermore, a number of evaluation criteria are being considered in order to achieve a more accurate inpainted result as well as a photorealistic representation of facial expressions and eye gaze direction under HMD occlusion.      

Current Status of the Project

Review on Facial Expression Reconstruction
100%
Data Collection and Building 3D Facial models
100%
Prepare a Research Article of the Collected Data Entitled: HEADSET (Human Emotion Awareness under Partial Occlusions Multimodal DataSet)
100%
Test and Modify Learning-Based Methods for Video Inpainting on HEADSET
100%
Improve the Quality of Facial Expression Reconstructions
100%
Design an Expression-aware video inpainting for HMD removal in XR applications
100%
Develop a Realistic Landmark-Guided Facial Video Inpainting Based on GANs
100%
3D Occlusion-Aware Video Inpainting
10%

Publications

  • HEADSET: Human Emotion Awareness under Partial Occlusions Multimodal DataSET (Published in: IEEE Transactions on Visualization and Computer Graphics- ISMAR 2023)
  • Expression-aware video inpainting for HMD removal in XR applications (Accepted in CVMP 2023)
  • Towards Realistic Landmark-Guided Facial Video Inpainting Based on GANs (Electronic Imaging Conference 2024)

Videos

Contact Information

Fatemeh Ghorbani
Ph.D. Student at TU Berlin
Department of Telecommunication Systems
Communication Systems Group
E-N1 Einsteinufer 17
10587 Berlin, Germany
Room: EN 306
Email addressghorbani.lohesara@tu-berlin.de

Find Fatemeh on Social Media: