Structure decoding commented out

[GregorKobsik]

Why did you comment out the structure decoding?

Is it not the idea of the paper to decode the geometry AND the structure independently?

DSG-Net/code/model_dsmnet_full.py

Lines 648 to 649 in 6c6d117

	# decode the current part box
	# box = self.box_decoder(node_struct_latent)

DSG-Net/code/model_dsmnet_full.py

Line 655 in 6c6d117

# ret.set_from_box_quat(box.view(-1))

DSG-Net/code/model_dsmnet_full.py

Line 702 in 6c6d117

# ret.set_from_box_quat(box.view(-1))

[GregorKobsik]

To fix the issue, I added the commented lines of code back. And also changed the used dataset in eval.py from the training set to the test set.

Can you confirm, that the pretrained model exp_vae_chair-full, would produce such results (original & reconstructed)? Is the full model the main model from your paper which you benchmarked against?

Shape No. 2197

Shape No. 2250