Evaluation on the Pix3D Dataset

[Sand-Glass]

In section 4.8 of the paper,

In detail, following [7, 8], we generate the training set using the data from the category of Chair in ShapeNet and synthesize images [35] with random background from the SUN database [36].

Could you release this training set and training code?

[LiyingCV]

Hi, sorry that we currently do not plan to release the training set. For generating the dataset, you can follow the generation approach referred to Pix2Vox++. For the training code, you can refer to the train.py.