Replicator semantics function not working properly

The function omni.replicator.core.modify.semantics is now working properly.

If I pass a single tuple like rep.modify.semantics([("class", "myclass")]) it works correctly and I can see the results when I render the semantic segmentation. But if I try to set more classes like rep.modify.semantics([("class", "myclass"), ("class2", "myotherclass")]) I can’t see the class2 when I render the semantic segmentation.

Moreover the doc at API — omni_replicator 1.6.4 documentation is wrong since it says Modify the orientation of the prims specified in input_prims to look at the specified target. and describe a not existing parameter target – The target to orient towards. If multiple prims are set, the target point will be the mean of their positions.

Hi @federico.domeniconi. Replicator questions are best asked in the SDG forum. I’ve moved this one over to that forum.

Hello,

Thank you for finding our embarrassing documentation copy & paste error, I’ll get that fixed ASAP.

As for the first part of your problem, I assume you are using BasicWriter?

In the documentation for BasicWriter there is a small line under the semantic_types parameter:

List of semantic types to consider when filtering annotator data. Default: [“class”]

if using BasicWriter add semantic_types to output the additional semanticTypes

writer = rep.WriterRegistry.get("BasicWriter")
writer.initialize(
    output_dir="semantics_classes",
    rgb=True,
    semantic_segmentation=True,
    semantic_types=['class', 'class2', 'class3']
)

This will output semantic data for class:xxx, class2:xxxx, and class3:xxxx labeled objects.

1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.