luigi-palladino / linemod-generator

Project for Computer Vision course of Professor Castellani @ Master's Degree in Computer Engineering for Robotics and Smart Industry @ UniVR 2022

linemod-generator

Project for Computer Vision course of Professor Castellani @ Master's Degree in Computer Engineering for Robotics and Smart Industry @ UniVR 2022

Generate 6D Pose Synthetic Dataset in Linemod format using Blender. Designed to be used with EfficientPose. The code produces rgb images with corresponding 6D Pose and depth images. Intrinsic matrix can also be set and printed from blender.

Refer to slides.pdf for further details.

Requirements:

• Python 3.7 recommended.
• Blender 3.2 required

Installation:

• Create a new environment with venv and use this command to install package requirements: python -m pip install -r requirements
• Go to Blender python interactor usually located at: Blender\3.2\python\bin\
• Copy paste this script and install pip executing it in a terminal located in Blender-python folder:
  • ./python.exe get_pip.py
• Install bpycv following installation instructions.
• Execute in Blender folder:

./blender -b --python-expr "from subprocess import sys,call;call([sys.executable]+'-m pip install -U tqdm'.split())"

Execution:

• Modifiy PATHS in linemod_generator.blend
• Execute linemod_generator.blend to generate 6D poses of the object
• Modifiy PATHS in unique_pipeline.ipynb
• Execute unique_pipeline.ipynb to convert in linemod dataset format

For problems write an e-mail @ [e-mail]