Please cite this work if you make use of our system in any of your own endeavors:
- SceneNet RGB-D: Can 5M Synthetic Images Beat Generic ImageNet Pre-training on Indoor Segmentation?, J. McCormac, A. Handa, S. Leutenegger, and A. J. Davison, ICCV '17
This is the SceneNet modified by BKAUTO to generate trinocular images. The build & compile steps are basically the same as in the original SceneNet RGBD. Codes in the third part (render part) are modified to provide correct depth image and multiple-eye views. Models and their appearance probability can be adjust in '/scene_generator/textfiles/'.
To better understanding the original design of SceneNet RGBD, I strongly recommend you to read the paper below before any further modification.
- SceneNet RGB-D: 5M Photorealistic Images of Synthetic Indoor Trajectories with Ground Truth, J. McCormac, A. Handa, S. Leutenegger, and A. J. Davison, arXiv '16