OpenAI’s contrastive image and text model (Radford and others, 2021). It puts images and text in one embedding space, which is how most open-vocabulary maps attach language to objects.
Mentioned in
14 papers:
- LOST-3DSG (2026)
- DAAAM (2026)
- TrackGraph (2026)
- BiMoSG (2026)
- 3DSG survey (2026)
- MERLION (2025)
- FunGraph (2025)
- ROMAN (2024)
- OpenGraph (2024)
- Open3DSG (2024)
- Clio (2024)
- Khronos (2024)
- HOV-SG (2024)
- ConceptGraphs (2023)
Notes: