ProjectionGAN

Last updated: 01/25/2022

This is my Bachelor of Engineering Thesis, (2019~2023, Schoolg of Art & Technology, Chung-Ang University.)

History

[Link] (Contest) 2021 CAU COMPASS - LINC+ CAPSTONE DESIGN AWARDS TOP 15 (Engineering and Natural Sciences PART) , Chung-Ang University, 2021. (12.28 Oral)

[Paper] (Domestic Conference) Real-time Depth Map Style Transfer in Object-Attention Region for Projection Mapping, Proceedings Vol. 5 No.2 pp5-8, NCISS-Winter, 2021. (12.29 Poster)

---

ProjectionGAN is Real-time Semantic Depth Style Transfer Model for Projection Mapping to Movable Object.

• How to do Depth Segmentation
• How to do Depth Style Trasnfer
• How to do DCamera-Projector Calibration

Implementation Details

Experimental Result

Getting Started ProjectionGAN

Training Environment

• Windows10
• Pytorch
• GPU with CC >= 3.0 : Compute capability (version)
• Python 3.8 or later with all requirements.txt dependencies installed.

(conda) #recommended
conda env create -f projection_gan.yml

(pip)
pip install -r requirements.txt

Related Project

CycleGAN | pix2pix-Torch