I would like to generate training-data for an image classifier, based on a CAD model. Therefore I would like to position my object in the center of the scene and render it from different positions.

Also there is an even bigger object (icosphere) placed in the center and each face of this ecosphere is equipped with a camera. The focus is on the center, where the CAD model is located.

Is there any possibility to implement this topic in code form and gain the desired images from different perspectives?

The results of this approach can be found the following paper:


I value your feedback, so let me know what you think!

All the best and stay healthy



Browse other questions tagged .