BlenderProc

TL;DR

BlenderProc is a modular, extendable procedural pipeline built on Blender for generating realistic images to train neural networks across various computer vision tasks.

Contribution

It introduces a user-friendly, modular pipeline with standard and extendable modules for synthetic image generation in Blender.

Findings

Provides realistic images for neural network training

Supports multiple computer vision tasks like segmentation and depth estimation

Easy to extend with custom modules

Abstract

BlenderProc is a modular procedural pipeline, which helps in generating real looking images for the training of convolutional neural networks. These can be used in a variety of use cases including segmentation, depth, normal and pose estimation and many others. A key feature of our extension of blender is the simple to use modular pipeline, which was designed to be easily extendable. By offering standard modules, which cover a variety of scenarios, we provide a starting point on which new modules can be created.