ScanRefer

From the catalogue. Listed by one researcher from one source page.

• id: scanrefer
• kind: language and reasoning annotation
• robot or device: 800 ScanNet scenes (class: sensor only)
• how collected: crowd annotation
• size as stated: 51,583 descriptions of 11,046 objects
• hours: not stated | hours only claimed: none | episodes: not stated
• year: 2019
• organisation: not stated
• licence: not stated (class: not stated)
• access: on request (class: on request)
• link (paper): https://arxiv.org/abs/1912.08830
• note: 3D object localization by language; access via ScanNet terms (not verified).

Confirm the licence at the link before relying on it.