use semantic segmentation for inference on a new dataset without labels about dgcnn.pytorch HOT 3 CLOSED

I think you can remove the prediction evaluation part of the code. Just only output the prediction results.

from dgcnn.pytorch.

Thank you Antao. Can you be more specific? Do you mean in def test() remove this section seg_pred = model(data) and lines related to pred ?
But before that when we read the data here test_loader = DataLoader(S3DIS(partition='test', ..., it expects labels, should I change the code to skip the label as input? Then where the inferred labels are stored?

from dgcnn.pytorch.

seg_pred = model(data) is the prediction results, not dataset labels.

I mean you can skip all label-related parts.

from dgcnn.pytorch.