imagenet-labels#

Code	IN1K.001
Validator	latest+
Compatibility	core-usd
Tags	✅

Summary#

Every SemanticsLabelsAPI:imagenet_1k label value must be a member of the ImageNet-1K taxonomy.

Description#

This requirement validates semantic label values against the ImageNet-1K vocabulary (1000 classes). It is a closed-vocabulary check: a value that is not a ImageNet-1K class fails. Matching is case-insensitive and tolerant of spacing (_/- are treated as spaces) and of a class’s display name or aliases — a value that matches only after that normalization passes with a warning that suggests the canonical class name.

This rule is layered on top of the vendor-neutral existence requirement SL.001: SL.001 asks whether a label exists; this rule asks whether the label is a valid ImageNet-1K class. It applies only to prims that opt into the taxonomy by applying its SemanticsLabelsAPI:imagenet_1k instance; prims without it are not affected.

Scope#

Evaluated over the stage’s defaultPrim and its descendants. The taxonomy is identified by the SemanticsLabelsAPI instance name imagenet_1k. Multiple taxonomies can coexist on the same prim (the schema is a “Multiple Apply” API), so this rule never conflicts with other taxonomy or Q-code labels.

Why is it required?#

Ground-truth and ML-training pipelines that consume this dataset’s label space need values drawn from that exact vocabulary.
Catches typos and out-of-vocabulary classes before the asset reaches a training run.

Examples#

# Valid: canonical ImageNet-1K classes
def Xform "Thing" (
    prepend apiSchemas = ["SemanticsLabelsAPI:imagenet_1k"]
)
{
    token[] semantics:labels:imagenet_1k = ["tench", "goldfish", "tiger shark"]
}

# Warning: matches after normalization -- suggests the canonical spelling "tench"
def Xform "ThingB" (
    prepend apiSchemas = ["SemanticsLabelsAPI:imagenet_1k"]
)
{
    token[] semantics:labels:imagenet_1k = ["Tench"]
}

# Invalid: not a ImageNet-1K class
def Xform "ThingC" (
    prepend apiSchemas = ["SemanticsLabelsAPI:imagenet_1k"]
)
{
    token[] semantics:labels:imagenet_1k = ["spaceship"]
}

How to comply#

Use class names from the ImageNet-1K taxonomy for semantics:labels:imagenet_1k values.
Prefer the canonical lowercase class name to avoid the non-canonical warning.