117 lines
6.1 KiB
YAML
117 lines
6.1 KiB
YAML
# ==============================================================================
|
|
# LinkML Slot Definition: has_object
|
|
# ==============================================================================
|
|
# Object classes detected in an image by a computer vision model.
|
|
#
|
|
# ONTOLOGY ALIGNMENT (verified against data/ontology/):
|
|
#
|
|
# | Ontology | Property / Class | File/Line | Mapping | Notes |
|
|
# |----------------|------------------------|------------------------------|----------|------------------------------------------------------------------------------------------|
|
|
# | **CIDOC-CRM** | `crm:P138_represents` | CIDOC_CRM_v7.1.3.rdf:4155 | related | E36 Visual Item→E1 CRM Entity. "Visually represents." Our slot stores ML output labels, |
|
|
# | | | | | not entity references, and records automated detection rather than human assertion. |
|
|
# | **CIDOC-CRM** | `crm:P62_depicts` | CIDOC_CRM_v7.1.3.rdf:2649 | related | E24 Physical Human-Made Thing→E1 CRM Entity. Shortcut of P65→P138. Same distinction: |
|
|
# | | | | | depicts is human-asserted and entity-valued; our slot is ML-asserted and string-valued. |
|
|
# | **OA** | `oa:hasBody` | oa.ttl:229-233 | related | Annotation→body resource. In tagging workflows (oa:tagging motivation), the body carries |
|
|
# | | | | | the tag/label. Structurally analogous but our slot flattens to a string list. |
|
|
#
|
|
# CREATED: 2026-02-10
|
|
# UPDATED: 2026-02-10
|
|
# ==============================================================================
|
|
|
|
id: https://nde.nl/ontology/hc/slot/has_object
|
|
name: has_object
|
|
title: Has Object
|
|
prefixes:
|
|
linkml: https://w3id.org/linkml/
|
|
hc: https://nde.nl/ontology/hc/
|
|
crm: http://www.cidoc-crm.org/cidoc-crm/
|
|
oa: http://www.w3.org/ns/oa#
|
|
imports:
|
|
- linkml:types
|
|
default_prefix: hc
|
|
slots:
|
|
has_object:
|
|
slot_uri: hc:hasObject
|
|
description: >-
|
|
Object class labels detected in an image or visual resource by a computer
|
|
vision model. Each value is a category name from the vocabulary used
|
|
during model training, such as COCO (80 common object categories),
|
|
ImageNet (1000 categories), or a custom heritage-specific taxonomy.
|
|
Multiple values indicate multiple distinct object classes were recognised
|
|
in the same image.
|
|
alt_descriptions:
|
|
nl: >-
|
|
Objectklassen die door een computervisiemodel in een afbeelding zijn
|
|
gedetecteerd, uitgedrukt als categorienamen uit het trainingsvocabularium.
|
|
de: >-
|
|
Von einem Computer-Vision-Modell in einem Bild erkannte Objektklassen,
|
|
ausgedrückt als Kategorienamen aus dem Trainingsvokabular.
|
|
fr: >-
|
|
Classes d'objets détectées dans une image par un modèle de vision par
|
|
ordinateur, exprimées sous forme de noms de catégories issus du
|
|
vocabulaire d'entraînement.
|
|
ar: >-
|
|
فئات الكائنات المكتشفة في صورة بواسطة نموذج رؤية حاسوبية، معبراً عنها
|
|
بأسماء فئات من مفردات التدريب.
|
|
id: >-
|
|
Kelas objek yang terdeteksi dalam gambar oleh model computer vision,
|
|
dinyatakan sebagai nama kategori dari kosakata pelatihan.
|
|
zh: >-
|
|
计算机视觉模型在图像中检测到的对象类别，以训练词汇表中的类别名称表示。
|
|
es: >-
|
|
Clases de objetos detectadas en una imagen por un modelo de visión
|
|
artificial, expresadas como nombres de categorías del vocabulario de
|
|
entrenamiento.
|
|
structured_aliases:
|
|
- literal_form: gedetecteerd object
|
|
predicate: EXACT_SYNONYM
|
|
in_language: nl
|
|
- literal_form: erkanntes Objekt
|
|
predicate: EXACT_SYNONYM
|
|
in_language: de
|
|
- literal_form: objet détecté
|
|
predicate: EXACT_SYNONYM
|
|
in_language: fr
|
|
- literal_form: كائن مكتشف
|
|
predicate: EXACT_SYNONYM
|
|
in_language: ar
|
|
- literal_form: objek terdeteksi
|
|
predicate: EXACT_SYNONYM
|
|
in_language: id
|
|
- literal_form: 检测到的对象
|
|
predicate: EXACT_SYNONYM
|
|
in_language: zh
|
|
- literal_form: objeto detectado
|
|
predicate: EXACT_SYNONYM
|
|
in_language: es
|
|
range: string
|
|
multivalued: true
|
|
related_mappings:
|
|
- crm:P138_represents # CIDOC_CRM_v7.1.3.rdf:4155-4169 - "represents" what a Visual Item visually depicts. Entity-valued and human-asserted, unlike our ML string labels.
|
|
- crm:P62_depicts # CIDOC_CRM_v7.1.3.rdf:2649-2661 - Shortcut "depicts" (Physical Human-Made Thing→CRM Entity). Same distinction as P138.
|
|
- oa:hasBody # oa.ttl:229-233 - Annotation body. In oa:tagging workflows the body carries the tag label; structurally analogous but our slot flattens to a string list.
|
|
aliases:
|
|
- object_classes_detected
|
|
examples:
|
|
- value: "person"
|
|
description: >-
|
|
COCO category detected in a museum surveillance image.
|
|
- value: "painting"
|
|
description: >-
|
|
Custom heritage-specific category detected in a gallery photograph.
|
|
- value: "vase"
|
|
description: >-
|
|
COCO/heritage category for a ceramic object detected in a collection image.
|
|
annotations:
|
|
custodian_types: '["*"]'
|
|
comments:
|
|
- >-
|
|
The vocabulary of valid values depends on the model used: COCO provides
|
|
80 common object categories, ImageNet provides 1000, and heritage
|
|
institutions may define custom taxonomies for domain-specific objects
|
|
such as paintings, manuscripts, pottery, or architectural elements.
|
|
- >-
|
|
This slot stores the raw string label output from the detection model.
|
|
For structured annotations with confidence scores, bounding boxes, or
|
|
provenance, consider using the Web Annotation (OA) data model with
|
|
an oa:tagging motivation instead.
|