depth-estimation-deep-learning

Fringe Pattern Projection with a Digital Twin in Blender

In this project we propose the implementation of a digital twin to simulate the reconstruction of 3D objects with the digital fringe projection technique. The main utility of this system is the flexibility to acquire the fringe images, obtaining a large number of images (faithful and with the same parameters of the real system) in a short time taking advantage of the GPU of the computer and thus being able to create feasible Datasets for the training of Deep Learning models.

Dependencies

• Blender
• Python3
• NumPy
• OpenCV
• Projector Add-on for Blender

*All necessary dependencies and software are Open-Sourse

Set-up and Installation

1. Download and install the Blender Projector Add-on from this repository. Guide to installing plugins in Blender
2. If you have CUDA-compatible graphics, you can select it for rendering in the following settings. *Blender Preferences > System > Cycles Render Devices > CUDA.
3. Install OpenCV and Scipy in Blender. Aquí recomiendo instalar los paquetes en tu instalación de Python y luego copiarlos en la ruta C:/Program Files/Blender Foundation/Blender 3.x/3.x/python/lib/site-packages/