User Tools

Site Tools


6a.001.TUT - Automatic Face Completion and Restoration

Project - Summary

Facial image de-occlusion refers to removing unnecessary occluded parts (e.g. scarf, glasses, cover) from a facial image, and then reconstructing its corresponding contents to a complete and realistic facial image conditioning on pre-defined image sets (T. Hosoi). Image in-painting proposed by Bertalmio et al. firstly (M. Bertalmio), also known as image completion, familiarly means to synthesize the missing pixels or remove unwanted pixels through learning context features from their surrounding regions, the whole image content, or external databases. In this project, we proposed an unsupervised method based on optimized Deep Convolutional Generative Adversarial Networks (DCGAN) (A. Radford) to address facial image de-occlusion and in- painting simultaneously in an iterative way.

Project - Team

Team Member Role Email Phone Number Academic Site/IAB
Moncef Gabbouj PI +358 (400) 736613 Tampere University
Jenni Raitoharju Co-PI +358 50 447 8418 Tampere University
Honglei Zhang Co-PI Not available Not available Tampere University
Lei Xu Researcher Not available Not available Tampere University
Juhani Ahonen Project Mentor Not available Not available Nokia - FI/Tampere

Project - Deliverables

1 Dataset of images and ground-truths for image-to-image mapping
2 Trained model for image mapping that is visually plausible to a human observer
3 Trained model for image mapping, enhancing recognition performance
4 Real-time implementation of both approaches

Project - Documents

FilenameFilesizeLast modified
6a.001.tut_poster_ppt.pptx5.8 MiB2019/08/14 15:25
6a.001.tut_executive_summary.pdf78.8 KiB2019/08/14 15:25
6a.001.tut_final_report.docx1.1 MiB2019/08/14 15:25
6a.001.tut_automatic_face_completion_and_restoration_poster_2017_fall_meeting.pptx779.5 KiB2019/08/14 15:25
6a.001.tut_poster_pdf.pdf370.8 KiB2019/08/14 15:25
6a.001.tut_ppt_presentation.pptx564.6 KiB2019/08/14 15:25
6a.001.tut_cvdi_mid-year_report.pdf274.3 KiB2019/08/14 15:25
6a.001.tut_confluence_project_page.pdf147.5 KiB2019/08/14 15:25
projects/year6/6a.001.tut.txt · Last modified: 2021/06/02 14:43 by sally.johnson