Skip to content

A procedural Blender pipeline for photorealistic training image generation

License

Notifications You must be signed in to change notification settings

gaokai15/BlenderProc

 
 

Repository files navigation

BlenderProc

This repository, forked from BlenderProc, is used to synthesize a desktop scenes dataset with a set of YCB objects and generate BOP data and COCO data.

BlenderProc rerun image

Usage

$ python3 config_generator.py [number of objects in each scene] [number of scenes] [output directory]

By default, the BOP data and COCO data will be stored at "/Path/to/BlenderProc/output"

Example:

$ cd BlenderProc
$ python3 config_generator.py 2 2 ./temp/output

The files will be stored at "/Path/to/BlenderProc/temp/output".

After the data is generated, the output folder structure is as below:

.
├── bop_data
│   ├── camera.json
│   ├── train_pbr
│   |   ├── 000000
│   |   |   ├── scene_camera.json
│   |   |   ├── scene_gt.json
│   |   |   ├── depth
│   |   |   |   ├── 000000.png
│   |   |   |   ├── 000001.png
│   |   |   |   ├── ...
│   |   |   ├── rgb
│   |   |   |   ├── 000000.png
│   |   |   |   ├── 000001.png
│   |   |   |   ├── ...
│   |   ├── 000001
│   |   ├── ...
├── coco_data
│   ├── coco_annotations.json
│   ├── rgb_0000.png
│   ├── rgb_0001.png
│   ├── ...

About

A procedural Blender pipeline for photorealistic training image generation

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 100.0%